Vehicle control device

The vehicle control device enhances steering feel by detecting vehicle balance changes and executing rapid learning adjustments, distinguishing them from disturbances to improve steering responsiveness.

JP2024168177AActive Publication Date: 2024-12-05SUZUKI MOTOR CORP
View PDF 4 Cites 0 Cited by

Patent Information

Application Number
JP2023084626
Authority / Receiving Office
JP · JP
Patent Type
Applications
Current Assignee / Owner
Filing Date
2023-05-23
Publication Date
2024-12-05

AI Technical Summary

Technical Problem

The midpoint position of the steering angle changes due to variations in vehicle balance, including occupant arrangement and loading, necessitating quick learning adjustments, but uniform learning conditions lead to sensitivity to disturbances like crosswinds, deteriorating steering feel.

Method used

A vehicle control device with a midpoint position learning mechanism that detects vehicle balance changes and executes learning at a higher speed when balance shifts, distinguishing it from disturbances, thereby improving steering feel.

Benefits of technology

Quickly adapts to vehicle balance changes while suppressing disturbances, enhancing steering feel by prioritizing learning based on vehicle balance adjustments.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 2024168177000001_ABST
    Figure 2024168177000001_ABST
Patent Text Reader

Abstract

To improve steering feeling by quickly executing learning for a change in vehicle balance while preventing the influence of disturbance on midpoint learning.SOLUTION: A vehicle control device includes: midpoint position learning means that executes midpoint learning for learning a midpoint position θctr of a steering angle; and vehicle balance change detection means that detects a change in vehicle balance. When the change in the vehicle balance is detected (S103), the midpoint learning (first midpoint learning) for the change in the vehicle balance is executed at a learning speed higher than midpoint learning (second midpoint learning) before the detection (S106).SELECTED DRAWING: Figure 3
Need to check novelty before this filing date? Find Prior Art

Description

[Technical field]

[0001] The present invention relates to a vehicle control device. [Background technology]

[0002] There are vehicle control devices that learn changes that occur in the midpoint position of the steering angle and correct the midpoint position. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] JP 2014-240234 A Summary of the Invention [Problem to be solved by the invention]

[0004] The midpoint position of the steering angle changes due to differences in individual characteristics of the vehicle, changes over time, etc., as well as changes in the vehicle balance.

[0005] One of the main factors that affect the vehicle's weight balance, that is, the change in the center of gravity, is the change in the location of the occupants (including the driver and passengers) inside the vehicle. For example, the vehicle's weight balance differs when there are occupants in the driver's seat and the passenger seat from when there are occupants in the driver's seat and the rear seat behind the driver's seat. In addition, changes in the weight balance can also occur when luggage is loaded in a biased position to the left or right.

[0006] In order to quickly reflect the effects of changes in vehicle balance in the midpoint position of the steering angle, it is necessary to learn the changes that have occurred in the midpoint position and to set the conditions for executing midpoint learning, which corrects the midpoint position, to be gentle in order to encourage learning.

[0007] However, if the conditions for executing learning were relaxed uniformly regardless of the cause of the change in the midpoint position, the learning would respond sensitively to temporary changes in the midpoint position caused by disturbances such as crosswinds, etc. This would cause the influence of the disturbances to be reflected in the midpoint position, which could worsen the steering feeling.

[0008] In view of this situation, the present invention aims to provide a vehicle control device that can improve steering feeling by quickly learning in response to changes in vehicle balance while suppressing the effect of external disturbances such as crosswinds on midpoint learning. [Means for solving the problem]

[0009] In order to solve the above-mentioned problems, a vehicle control device according to one embodiment of the present invention includes a midpoint position learning means that performs midpoint learning to learn the midpoint position of the steering angle, and a vehicle balance change detection means that detects a change in vehicle balance, and when the vehicle balance change detection means detects a change in the vehicle balance, the midpoint position learning means performs the midpoint learning for the change in vehicle balance at a learning speed that is faster than the midpoint learning before the detection. Effect of the Invention

[0010] According to one aspect of the present invention, when a change occurs in the vehicle balance and the change is detected, midpoint learning for the change in the vehicle balance is executed, and the midpoint learning in this case is executed at a learning speed faster than that of the midpoint learning before the change is detected. This makes it possible to quickly execute learning for the change in the balance of the vehicle itself, actively reflect the effect of the change in the midpoint position of the steering angle, and improve the steering feeling, while suppressing the effect of disturbances such as crosswinds on the learning of the midpoint position, thereby preventing the steering feeling from deteriorating. [Brief description of the drawings]

[0011] [Figure 1] 1 is a schematic diagram showing a configuration of a vehicle control device according to an embodiment of the present invention; [Diagram 2] FIG. 2 is a schematic diagram showing the internal configuration of an EPS controller. [Diagram 3] 4 is a flowchart showing an overall flow of a midpoint position learning control. [Figure 4] 13 is a flowchart showing the contents of a first midpoint learning process. [Diagram 5] 13 is a flowchart showing the contents of a second midpoint learning process. [Figure 6] FIG. 1 is an explanatory diagram conceptually showing a method for evaluating vehicle balance according to a seat belt fastening state. [Figure 7] 4 is a time chart showing a change in a learned value CL of the midpoint position due to midpoint position learning control. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0012] Hereinafter, an embodiment of the present invention will be described with reference to the drawings.

[0013] (Steering control system configuration) FIG. 1 is a schematic diagram showing the configuration of a vehicle steering control system S according to one embodiment of the present invention.

[0014] The steering control system S includes an electric power steering device 1 and a controller (hereinafter referred to as "EPS controller") 101 that controls the operation of the electric power steering device 1. The EPS controller 101 constitutes a "vehicle control device" according to this embodiment.

[0015] In this embodiment, a vehicle is targeted in which a steering wheel operated by a driver and wheels to be operated (i.e., steered wheels) are mechanically connected and configured to be capable of coordinated operation. In such a vehicle, the operation torque applied to the steering wheel is transmitted to the left and right steered wheels via a steering shaft extending from the steering wheel and a power conversion device such as a rack and pinion mechanism. The electric power steering device 1 includes an electric motor, and assists the steering operation performed by the driver via the steering wheel with the electric motor. Specifically, an assist torque according to the steering operation by the driver is generated by the electric motor, and this assist torque is added to the operation torque applied by the driver to the steering shaft. It is not limited to the electric power steering device 1, and it is also possible to adopt a hydraulic power steering device. The steering control system S may be of a so-called steering-by-wire type in which the steering wheel and the steered wheels are not mechanically connected to each other but are separated from each other.

[0016] The steering control system S is equipped with a steering angle sensor 201, a wheel speed sensor 202, a yaw rate sensor 203, a lateral acceleration sensor 204, and a seat belt reminder switch 205.

[0017] The steering angle sensor 201 detects the rotational position of the steering wheel as the steering angle θstr of the vehicle. The wheel speed sensor 202 detects the rotational speed WSP of the wheel to be detected. In this embodiment, the pair of left and right front wheels (right front wheel, left front wheel) and the pair of left and right rear wheels (right rear wheel, left rear wheel) are each detected. Based on the wheel speed WSP detected by the wheel speed sensor 202, it is possible to detect the traveling speed (i.e., vehicle speed) VSP of the vehicle. The yaw rate sensor 203 detects the rotational angular velocity (i.e., yaw rate) Ryaw of the vehicle about the vertical axis. The lateral acceleration sensor 204 detects the acceleration ACCl acting in the lateral direction with respect to the vehicle. The seat belt reminder switch 205 is configured to be able to output a signal corresponding to whether the seat belt is fastened or not, and is provided for each seat where the seat belt is installed. The output signals of these various sensors 201 to 204 and the switch 205 are input to the EPS controller 101. Upon receiving input of various sensor information, the EPS controller 101 executes a predetermined calculation related to the steering control of the vehicle, and outputs a control signal to the electric power steering device 1 according to the result of the calculation.

[0018] (EPS controller internal structure) FIG. 2 is a schematic diagram showing the internal configuration of the EPS controller 101.

[0019] The EPS controller 101 is roughly divided into a vehicle balance change detection unit 111, a straight / stopped vehicle determination unit 112, and a midpoint position learning unit 113. In this embodiment, the EPS controller 101 is configured as an electronic control unit, and is configured by a microcontroller having a central processing unit (CPU), various memory units such as ROM and RAM, an input / output interface, etc.

[0020] The vehicle balance change detection unit 111 detects a change in vehicle balance when it occurs. A change in vehicle balance occurs due to a change in the weight balance of the vehicle and other factors. A change in weight balance occurs due to a change in the position of occupants (passengers including the driver and passengers) in the vehicle, and also occurs when luggage is loaded in a state where it is biased to one side or the other. Other than a change in weight balance, factors that cause a change in vehicle balance include an imbalance in tire air pressure, a mismatch in tire type and performance, and the like. Specifically, a change in vehicle balance occurs when tire air pressure drops in some wheels or when a different type of tire (e.g., a temporary tire or a pre-tender tire) is attached to some wheels. In this embodiment, a change in weight balance due to a change in occupant position is detected based on an output signal from the seat belt reminder switch 205.

[0021] The straight / stopped vehicle determination unit 112 determines whether the vehicle is traveling straight or stopped. Specifically, based on the output signals of the steering angle sensor 201, the wheel speed sensor 202, the yaw rate sensor 203, and the lateral acceleration sensor 204, when the steering angle θstr is equal to or smaller than a predetermined value θ1, the yaw rate Ryaw is equal to or smaller than a predetermined value R1, and the lateral acceleration ACCl is equal to or smaller than a predetermined value A1, the vehicle is determined to be traveling straight. These predetermined values ​​θ1, R1, and A1 are set according to the vehicle speed VSP, and for example, the steering angle θstr is set to a smaller value as the vehicle speed VSP increases. Furthermore, the predetermined values ​​θ1, R1, and A1 are set to different values ​​when midpoint learning is performed for a change in vehicle balance and when midpoint learning is performed for other factors, and when midpoint learning is performed for a change in vehicle balance, the predetermined values ​​θ1, R1, and A1 are increased and switched to settings that encourage a determination that the vehicle is traveling straight. The vehicle is determined to be stopped when the vehicle speed VSP is equal to or less than a predetermined value V1. The vehicle speed VSP, the steering angle θstr, the yaw rate Ryaw, and the lateral acceleration ACCl are state parameters that correlate with the running state of the vehicle.

[0022] When a change occurs in the midpoint position or a deviation from the current midpoint position, the midpoint position learning unit 113 learns the change and executes midpoint learning to correct the midpoint position. Changes in the midpoint position occur due to changes in vehicle balance, as well as disturbances such as road cant and crosswinds. Furthermore, changes in the midpoint position occur due to differences in individual vehicle characteristics (for example, differences in parts alignment from the steering wheel to the steered wheels) and changes over time such as part deterioration. In this embodiment, changes due to changes in vehicle balance, differences in individual vehicle characteristics, changes over time, and the like are the subject of midpoint learning, while changes due to disturbances are considered to be temporary or transient and are not the subject of midpoint learning.

[0023] The midpoint position learning unit 113 learns the midpoint position at different speeds depending on whether the change in the midpoint position is due to a change in vehicle balance or due to a factor other than the change in vehicle balance, and performs midpoint learning when the change in the midpoint position is due to a change in vehicle balance at a higher learning speed than midpoint learning when the change is due to a factor other than the change in vehicle balance. In this embodiment, the "factors other than a change in vehicle balance" that cause a change in the midpoint position are differences in individual vehicle characteristics, aging, etc.

[0024] The midpoint position learning unit 113 includes a first learning unit 113a and a second learning unit 113b, and the first learning unit 113a quickly learns the midpoint position that has changed due to a change in vehicle balance at a relatively high learning speed. On the other hand, the second learning unit 113b slowly learns the midpoint position that has changed due to factors other than the change in vehicle balance, that is, differences in individual vehicle characteristics and aging, and takes time to learn the midpoint position at a learning speed slower than that of the first learning unit 113a.

[0025] The midpoint position learning unit 113 calculates a learning value CL of the midpoint position as a result of the midpoint learning. The midpoint position θctr can be corrected by adding this learning value CL to the initial value θ0 of the midpoint position (θctr=θ0+CL). In this embodiment, the initial value θ0 of the midpoint position is set to 0 (zero value). Even if the initial value θ0 is greater than 0 and there is a deviation in the midpoint position θctr of the vehicle at the time of shipment, this deviation can be eliminated by the midpoint learning executed by the second learning unit 113b as being due to differences in individual characteristics of the vehicle.

[0026] The EPS controller 101 calculates the difference Δθ (=θstr-θctr) between the steering angle θstr detected by the steering angle sensor 201 and the midpoint position θctr, and calculates the assist torque TRQs (=G×Δθ) by multiplying this difference Δθ by a predetermined gain G. Then, the EPS controller 101 outputs a command signal to the electric power steering device 1, causing the electric motor to generate the assist torque TRQs.

[0027] (EPS controller operation) The operation of the EPS controller 101 will be described below with reference to a flowchart.

[0028] 3 is a flowchart showing an overall flow of the midpoint position learning control. The ESP controller 101 repeatedly executes the midpoint position learning control according to the procedure shown in FIG.

[0029] In S101, output information from various sensors and switches such as the steering angle sensor 201, the wheel speed sensor 202, and the seat belt reminder switch 205 is read.

[0030] In S102, it is determined whether the vehicle is stopped. If the vehicle is stopped, the process proceeds to S103, and if the vehicle is not stopped, that is, if the vehicle is moving, the process proceeds to S105.

[0031] In S103, it is determined whether a change in vehicle balance has been detected. If a change in vehicle balance has been detected, the process proceeds to S104, and if not, the current control ends and the process waits until the vehicle starts. A change in vehicle balance is detected based on the output signal of the seat belt reminder switch 205. Here, detection of a change based on the output signal of the seat belt reminder switch 205 will be described with reference to FIG. 6.

[0032] FIG. 6 is an explanatory diagram conceptually showing a method for evaluating vehicle balance depending on the state of seat belt use.

[0033] FIG. 6(a) shows a seating arrangement for five people in two rows, front and rear, with the top side facing the vehicle's forward direction on the paper. With the forward direction facing forward, the first row is the front row and the second row is the rear row. In this arrangement, the seat on the left side of the first row (1A) corresponds to the driver's seat, and the seat on the right side of the first row (1C) corresponds to the passenger seat. In this embodiment, there is no seat available in the center of the first row (1B). The three seats (2A, 2B, 2C) in the second row correspond to the left rear seat, the center rear seat, and the right rear seat, respectively.

[0034] FIG. 6(b) shows reference values ​​i (i1 to i7) that indicate the arrangement of passengers in the vehicle, corresponding to positions A to C and the number of passengers N. In this embodiment, the reference value i indicates the degree of influence of the number of passengers N at each position A to C on the left-right weight ratio. For example, when only the driver is in the vehicle, the reference value i1 corresponding to the number of passengers N=1 at position A is read out and set as the reference value i (i=i1). When there is a passenger in the passenger seat in addition to the driver, the reference value i1 and the reference value i6 corresponding to the number of passengers N=1 at position C are read out and the sum of these values ​​is set as the reference value i (i=i1+i6). The presence of passengers in each seat can be determined based on the output signal of the seat belt reminder switch 205. In the same manner, the reference values ​​i1 to i7 corresponding to the number of passengers N for each of positions A to C are read out and summed up to obtain the reference value i that indicates the arrangement of passengers in the entire vehicle, and the weight balance can be roughly determined.

[0035] In this manner, in this embodiment, vehicle balance is evaluated based on the number and arrangement of seats in which occupants have their seat belts fastened, and if there is a change in the number or arrangement of seats in which the seat belts are fastened, this is detected as a change in vehicle balance.

[0036] Returning to Fig. 3, in S104, the flag FRG is set to 1. The flag FRG is set to indicate that, when a change in vehicle balance is detected, midpoint learning for the change in vehicle balance (hereinafter referred to as "first midpoint learning") is being executed. The flag FRG is normally set to 0, and after a change in vehicle balance is detected, it is set to 1 until the first midpoint learning is completed.

[0037] In S105, it is determined whether or not the flag FRG is 1. If the flag FRG is 1, it is determined that first midpoint learning is being executed, and the process proceeds to S106, and if the flag FRG is not 1, the process proceeds to S108.

[0038] In S106, first midpoint learning is started or continued. The first midpoint learning is learning for a change in the midpoint position θctr due to a change in the vehicle balance, and is based on a first learning condition. The first midpoint learning is executed by the first learning unit 113a according to the procedure shown in the flowchart of FIG.

[0039] In S107, it is determined whether or not the flag FRG is 0. If the flag FRG is 0, it is determined that the first midpoint learning is completed, and the process proceeds to S108, but if the flag FRG is not 0, it is determined that the first midpoint learning is not completed, and the current control ends.

[0040] In S108, the second midpoint learning is resumed or continued. The second midpoint learning is learning for changes in the midpoint position θctr due to factors other than changes in vehicle balance, specifically, differences in individual characteristics of the vehicle and changes over time, and is based on second learning conditions. The second midpoint learning is executed by the second learning unit 113b according to the procedure shown in the flowchart of FIG. 5.

[0041] FIG. 4 is a flowchart showing the contents of the first midpoint learning process.

[0042] In S201, the learning initial value CLi of the first midpoint learning to be executed this time (hereinafter referred to as the "current initial value") is set. n In this embodiment, a learning initial value CLi corresponding to the vehicle balance is set in advance and stored in a memory unit of the EPS controller 101. When the EPS controller 101 detects a change in the vehicle balance, it reads out the learning initial value CLi corresponding to the changed vehicle balance from the memory unit and sets the current initial value CLi n Set to.

[0043] In S202, the currently set learning value (hereinafter referred to as the "current learning value") CL, that is, the most recent learning value, is read, and the learning initial value (hereinafter referred to as the "previous initial value") CLi in the first midpoint learning executed last time is read. n-1 Load.

[0044] In S203, the current learning value CL and the previous initial value CLi n-1 Calculate the difference between this and the initial value CLi n The learning value CL is updated by adding it to the learning value CL. The update of the learning value CL in S203 is performed once every time a change in the vehicle balance is detected, and the first midpoint learning is started from the updated learning value CL. That is, in this embodiment, the updated learning value CL according to the following formula (1) is substantially the initial learning value. CL=CLi n +(CL-CLi n-1 ) …(1)

[0045] In S204, it is determined whether the vehicle is traveling straight. If the vehicle is traveling straight, the process proceeds to S205. If the vehicle is not traveling straight, the process returns to the flowchart shown in FIG. 3 and executes the process of S107. As described above, in the first midpoint learning, the thresholds θ1, R1, and A1 for determining whether the vehicle is traveling straight are set to relatively large values ​​to promote the learning by encouraging the determination that the vehicle is traveling straight. The thresholds θ1, R1, and A1 define the first learning condition.

[0046] In S205, it is determined whether the actual midpoint position θctr' calculated from the detection value of the steering angle sensor 201 and the calculated midpoint position θctr by learning are sufficiently close. In this embodiment, since the initial value θ0 of the midpoint position is set to 0 (zero value), the calculated midpoint position θctr is equal to the learning value CL (θctr=CL). The difference between the actual midpoint position θctr' and the learning value CL is calculated, and if it is equal to or less than a predetermined value SL, the process proceeds to S207, and if it is greater than the predetermined value SL, the process proceeds to S206, since further learning for the change in vehicle balance is required. The actual midpoint position θctr' can be calculated as the average value of the steering angle θstr detected by the steering angle sensor 201 (for example, a moving average value according to time or distance).

[0047] In S206, the difference between the actual midpoint position θctr' and the learning value CL is multiplied by a predetermined first learning coefficient Kc1, and the product is added to the current learning value CL to update the learning value CL. This gradually eliminates the difference between the actual midpoint position θctr' and the learning value CL, and the learning value CL approaches the actual midpoint position θctr'. The first learning coefficient Kc1, together with the threshold values ​​θ1, R1, and A1 for determining straight driving, defines the first learning condition, and is larger than the second learning coefficient Kc2 described later. CL = CL + Kc1 × (θctr' - CL) ... (2)

[0048] In S207, it is determined whether the first midpoint learning is stable. The stability of the first midpoint learning is determined by whether the amount of change in the learned value ΔCL (=Kc1×(θctr′−CL)) calculated in S206, that is, the amount of change in the learned value per learning, ΔCL, has decreased to a predetermined value or less. If the first midpoint learning is stable, the process proceeds to S208, and if it is not stable, the process returns to the flowchart shown in FIG. 3 and executes the process of S107.

[0049] In S208, it is determined that midpoint learning (first midpoint learning) for the currently detected change in vehicle balance has been completed, and the flag FRG is set to 0.

[0050] FIG. 5 is a flowchart showing the contents of the second midpoint learning process.

[0051] In S301, it is determined whether the vehicle is traveling straight. If the vehicle is traveling straight, the process proceeds to S302, and if the vehicle is not traveling straight, the process returns to the flowchart shown in FIG. 3 and ends this control. As described above, in the second midpoint learning, the thresholds θ1, R1, and A1 for determining whether the vehicle is traveling straight are set to values ​​smaller than those in the first midpoint learning, and the learning is suppressed by limiting the determination that the vehicle is traveling straight. The thresholds θ1, R1, and A1 define the second learning conditions.

[0052] In S302, similarly to S205, it is determined whether the actual midpoint position θctr' calculated from the detection value of the steering angle sensor 201 is sufficiently close to the midpoint position θctr (=CL) calculated by learning. If the difference between the actual midpoint position θctr' and the learned value CL is equal to or less than a predetermined value SL, the current control is terminated, and if it is greater than the predetermined value SL, the process proceeds to S303 in order to learn factors other than changes in vehicle balance.

[0053] In S303, the difference between the actual midpoint position θctr' and the learning value CL is multiplied by a predetermined second learning coefficient Kc2, and the product is added to the current learning value CL to update the learning value CL. The second learning coefficient Kc2 defines the second learning condition and is smaller than the first learning coefficient Kc1 described above. Under these conditions, the second midpoint learning proceeds over time at a slower learning speed than the first midpoint learning. CL = CL + Kc2 × (θctr' - CL) ... (3)

[0054] (Explanation of action and effect) When a change occurs in the vehicle balance, the midpoint position θctr of the steering angle changes due to the change. Therefore, when a change in the vehicle balance is detected, midpoint learning is performed to compensate for the change that occurs in the midpoint position θctr of the steering angle.

[0055] Midpoint learning for a change in vehicle balance (first midpoint learning) is performed at a higher learning speed than midpoint learning before the change is detected, i.e., midpoint learning for a factor other than the change in vehicle balance (second midpoint learning). In this embodiment, the learning speed is adjusted by setting the thresholds θ1, R1, and A1 for straight-line driving judgment and the learning coefficients Kc1 and Kc2. The frequency of midpoint learning can be adjusted by setting the thresholds θ1, R1, and A1, and the amount of change in the learning value per learning can be adjusted by setting the learning coefficients Kc1 and Kc2.

[0056] Specifically, by increasing the thresholds θ1, R1, and A1, the frequency of midpoint learning can be increased, and by increasing the learning coefficients Kc1 and Kc2, the amount of change in the learning value per learning can be increased. Either of these methods can have a substantial effect of promoting midpoint learning.

[0057] In the first midpoint learning for changes in vehicle balance, the threshold values ​​θ1, R1, A1 and learning coefficient Kc1 are set to larger values ​​than in the second midpoint learning for other factors, thereby facilitating rapid learning of changes in the balance of the vehicle itself (due to weight imbalance, imbalance in tire air pressure, mismatch in tire type and performance, etc.) and actively reflecting the effects of this in the steering angle midpoint position θctr, thereby making it possible to improve the steering feeling.

[0058] On the other hand, the influence of disturbances such as road cant and cross winds on the midpoint learning can be suppressed, and the influence of the disturbances on the midpoint position θctr of the steering angle can be suppressed to suppress a situation in which the steering feeling deteriorates.

[0059] FIG. 7 is a time chart showing changes in the learned value CL of the midpoint position due to the midpoint position learning control.

[0060] 7, it is assumed that a change in vehicle balance is detected at time t0. The change in vehicle balance is detected while the vehicle is stopped.

[0061] After detecting a change in the vehicle balance, if it is determined at time t11 that the vehicle is traveling straight ahead, the learning value CL is updated from the current learning value CL (=CL1) to start the first midpoint learning. The first update of the learning value CL after detecting a change in the vehicle balance is performed by updating the learning initial value CLi (=CLi n ) and is based on the above formula (1). If there is no change in vehicle balance and this change is not detected, the midpoint learning before the detection, that is, the second midpoint learning, is continued. In other words, when no change in vehicle balance is detected, the midpoint learning is a continuation of the midpoint learning from the previous time.

[0062] If the updated learned value CL is far from the midpoint position θctr' of the actual steering angle and the difference between them (=θctr'-CL) is greater than the threshold value SL, the first midpoint learning is continued on the condition that the vehicle is traveling straight. Specifically, the change amount ΔCLa in the learned value according to the difference is calculated using the above formula (2), and this is sequentially added to the learned value CL to update the learned value CL. The update of the learned value CL based on the difference is performed every cycle T1 as long as the vehicle continues to travel straight. The update cycle T1 of the learned value CL may be the same as or different from the execution cycle of the midpoint position learning control (the execution cycle of the control shown in FIG. 3). For example, the update cycle T1 can be set to a cycle longer than the execution routine of the midpoint position learning control.

[0063] As the learned value CL approaches the actual midpoint position θctr', the change amount ΔCLa in the learned value decreases, and when it reaches a predetermined value (time t1n), the first midpoint learning is deemed to have stabilized and is terminated, and the second midpoint learning is started or resumed with the learned value CL at that time (CL2=A+CLa). A=CLi n +(CL-CLi n-1 ) …(4.1) CLa=Σ(Kc1×(θctr'-CL)) …(4.2)

[0064] Here, the second midpoint learning compensates for changes in the midpoint position due to factors other than changes in vehicle balance, such as differences in individual vehicle characteristics and aging, and is performed over time at a low learning speed. Specifically, the update period T2 of the learned value CL is extended more than that of the first midpoint learning, the frequency of learning is reduced, and the change amount ΔCLb of the learned value per learning is reduced. The update period T2 can be extended by limiting the determination that the vehicle is traveling straight ahead by reducing the thresholds θ1, R1, and A1, or by setting the update period T2 separately from the execution period of the midpoint position learning control and setting it to a period longer than the update period T1 in the first midpoint learning.

[0065] In this way, by increasing the amount of change in the learning value per learning or increasing the frequency of learning in midpoint learning for changes in vehicle balance, the learning speed of midpoint learning for changes in vehicle balance can be easily adjusted, and a higher learning speed can be achieved in the first midpoint learning than in the second midpoint learning.

[0066] A change in the steering angle midpoint position θctr can be properly detected by a change in the weight balance of the vehicle, and the steering feeling can be improved through midpoint learning.

[0067] Here, the change in the midpoint position θctr due to a change in the weight balance is not temporary or transient, unlike changes due to disturbances such as road cant or crosswinds. Therefore, even if the learning speed is increased and learning is performed quickly without taking a long time, it is not of a nature that would cause a deterioration in steering feeling as premature learning.

[0068] Midpoint learning is performed on the condition that the vehicle is traveling straight, and when a change in vehicle balance is detected, the threshold values ​​θ1, R1, and A1 for determining straight-line driving are increased and the conditions for determining straight-line driving are relaxed. This promotes the execution of midpoint learning in response to changes in vehicle balance and quickly reflects the effects of these changes in the midpoint position of the steering angle, while suppressing the execution of midpoint learning in response to factors other than changes in vehicle balance, making it difficult for the effects of external disturbances to be reflected in the midpoint position of the steering angle.

[0069] After the midpoint learning begins, the learning is determined to have stabilized when the amount of change in the learned value per learning, ΔCL (ΔCLa), falls below a predetermined value. The thresholds θ1, R1, and A1, which were increased during the first midpoint learning, are then decreased again. This allows the midpoint learning to be continued unnecessarily under gentle conditions, thereby making it possible to avoid a situation in which a temporary change in the midpoint position due to a disturbance, etc., affects the learning.

[0070] Here, if the change amount ΔCL (ΔCLa) of the learned value per learning excessively increases after the midpoint learning is started, the thresholds θ1, R1, and A1 may be decreased from the value increased during the first midpoint learning, so that the condition for determining straight-line driving continues to be relaxed even under the influence of disturbances, etc., and the execution of midpoint learning is promoted, thereby making it possible to suppress a situation in which the influence of disturbances, etc. is reflected in the midpoint position. For example, a predetermined value larger than the threshold (S207) used for stabilization determination of learning is set, and if the change amount ΔCL of the learned value increases beyond this predetermined value, the thresholds θ1, R1, and A1 are decreased while the change amount ΔCL is maintained at a value larger than the predetermined value.

[0071] A learning initial value CLi according to the vehicle balance is preset, and when a change in the vehicle balance is detected, the learning initial value A=CLi according to the changed vehicle balance is set. n +(CL-CLi n-1 ), it is possible to shorten the time until learning is completed.

[0072] It is possible to appropriately and easily detect changes in vehicle balance due to an increase or decrease in the number of occupants or changes in their position depending on the seat belt fastening status. Furthermore, it is possible to appropriately evaluate the vehicle balance even when the vehicle is in a tilted state, such as when the vehicle is on a canted road or on an uphill or downhill road, and to appropriately detect any changes in the vehicle balance.

[0073] Changes in the balance of the vehicle itself generally occur when the vehicle is stopped and passengers get on and off, or luggage is loaded and unloaded, etc. Therefore, by detecting changes in vehicle balance when the vehicle is stopped, it is possible to prevent erroneous detection of changes in vehicle balance and erroneous learning of the midpoint position based on this, and to avoid an increase in the control load due to unnecessary judgments.

[0074] In this embodiment, after a change in vehicle balance is detected (time t0), the learned value CL is updated from the current learned value CL (=CL1) until it is determined that the vehicle is traveling straight ahead (time t11).

[0075] Alternatively, when a change in vehicle balance is detected, the learning value CL may be updated based on the change, and learning based on the difference between the actual midpoint position θctr' and the learning value CL may be performed while the vehicle is traveling straight. This allows the update of the learning value CL using the learning initial value CLi to be completed before the vehicle starts, and makes it possible to eliminate the discomfort caused by the change in the midpoint position during the period until it is determined that the vehicle is traveling straight (between time t0 and t11). Furthermore, when the change in vehicle balance is large, the correction of the midpoint position by learning is performed while the vehicle is traveling, making it possible to avoid a situation in which excessive fluctuations occur in the assist torque.

[0076] Furthermore, in this embodiment, when a change in the vehicle balance is detected, the learning initial value A=CLi n +(CL-CLi n-1 ) and start midpoint learning from the learning initial value A.

[0077] However, if a change in vehicle balance is detected but there is no significant change in the vehicle balance before and after the detection and no significant change in the midpoint position, the midpoint learning may be started from the final value of the midpoint learning before the detection (i.e., the current learned value CL), and the closer the changed vehicle balance is to the pre-change vehicle balance, the higher the learning speed may be. This makes it possible to quickly complete learning of slight changes in vehicle balance and quickly stabilize the steering feeling.

[0078] The steering control system S is not limited to a system that assists the driver in steering, but can also be applied to so-called self-driving cars that are configured to automatically control the steering angle of the vehicle. This is because in a self-driving car, it is necessary to grasp the midpoint position that serves as the reference for steering when driving the vehicle along a target route, and by learning the changes that occur in the midpoint position and correcting the midpoint position, it is possible to improve the driving stability of the self-driving car. [Explanation of symbols]

[0079] S...vehicle steering control system, 1...electric power steering device, 101...EPS controller, 201...steering angle sensor, 202...wheel speed sensor, 203...yaw rate sensor, 204...lateral acceleration sensor, 205...seat belt reminder switch.

Claims

1. a midpoint position learning means for executing midpoint learning for learning a midpoint position of a steering angle; A vehicle balance change detection means for detecting a change in vehicle balance, A vehicle control device, wherein when the vehicle balance change detection means detects a change in the vehicle balance, the midpoint position learning means performs the midpoint learning for the change in the vehicle balance at a learning speed higher than the midpoint learning before the detection.

2. 2. The vehicle control device according to claim 1, wherein the midpoint position learning means increases an amount of change in a learned value per learning or increases a frequency of learning in the midpoint learning in response to a change in the vehicle balance.

3. 2. The vehicle control device according to claim 1, wherein the vehicle balance change detection means detects a change in weight balance of the vehicle as the change in the vehicle balance.

4. The vehicle further includes a straight-line running determination means for determining that the vehicle is running straight when a state parameter correlated with a running state of the vehicle is within a predetermined range, the midpoint position learning means executes the midpoint learning on condition that the straight running determination means determines that the vehicle is running straight, 2. The vehicle control device according to claim 1, wherein the straight ahead running determination means, when the vehicle balance change detection means detects a change in the vehicle balance, expands the predetermined range to a value larger than that before the detection.

5. 5. The vehicle control device according to claim 4, wherein the straight-ahead running determination means, when an amount of change in the learning value per learning in the midpoint learning increases beyond a predetermined value after the midpoint learning in response to the change in the vehicle balance is started, narrows the predetermined range to a value smaller than that before the predetermined value was reached.

6. The midpoint position learning means a learning initial value of the midpoint position according to the vehicle balance; 2. The vehicle control device according to claim 1, wherein, when a change in the vehicle balance is detected, the midpoint learning is started from the learning initial value corresponding to the changed vehicle balance.

7. 7. The vehicle control device according to claim 6, wherein the vehicle balance change detection means sets the learning initial value of the midpoint position based on a state of fastening of a seat belt of a seat provided in the vehicle.

8. The midpoint position learning means when a change in the vehicle balance is detected, the midpoint learning is started from a final value of the midpoint learning before the detection; 2. The vehicle control device according to claim 1, wherein the midpoint learning is executed at a higher learning speed as the post-change vehicle balance is closer to the pre-change vehicle balance.

9. 9. The vehicle control device according to claim 1, wherein the vehicle balance change detection means detects the change in the vehicle balance when the vehicle is stopped.

Citation Information

Patent Citations

  • Steering support device

    JP2010120504A

  • Control device and control method

    JP2019151243A

  • Vehicle control device

    JP2022112211A

  • Steering control device

    JP2014240234A