JPH11310120A - Resonance gain operation device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To operate a resonance gain at a high accuracy regardless of the variation of a tire spring constant. SOLUTION: This operation device is provided with a brake torque detector 14 to detect the brake torque Tb; a wheel speed sensor 16 to detect the wheel speed signal ω; an α grade.tire spring constant operation part 12 to operate the grade α and the tire spring constant K to the slip speed of the friction coefficient μfrom the detected brake torque Tb and the wheel speed signal ω, depending on the resonance model of a vibration system which is composed of a car body, a wheel, and a road surface; a phase operation part 15 to operate the phase difference θd from the minute exciting vibration of the brake torque to the wheel speed minute vibration from α and K; and a gain operation part 17 to operate the amplitude of the minute vibration of the wheel speed ω having the phase difference θd and the resonance frequency of the above vibration system and the like. Even though the phase difference θd is changed by the variation of the tire spring constant, the resonance gain G4 responding to the change can be operated at a high accuracy constantly, by such a structure.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a resonance gain calculation device, and more particularly, to a time series data of a detected brake torque based on a resonance model of a vibration system composed of a vehicle body, wheels, and a road surface. And the time series data of the wheel speed, the gradient of the friction coefficient μ with respect to the slip speed (hereinafter referred to as
And the tire spring constant is estimated.
The present invention relates to a resonance gain calculation device that calculates a resonance gain with higher accuracy based on a gradient and a tire spring constant regardless of a change in tire air pressure.

[0002]

2. Description of the Related Art Conventionally, there has been proposed a technique for judging a slip state between a wheel and a road surface based on vibration characteristics of a vibration system composed of a vehicle body, a wheel, and a road surface (Japanese Patent Application No. Hei 7 (1995) -19783).
-220920).

According to one application of this technology, first,
The resonance frequency ω∞ of the vibration system during tire grip (described later)
(See formula (7)) to slightly excite the braking force (brake torque). Then, the characteristic calculating the resonant gain G _d is a small amplitude ratio of the wheel speed in the resonant frequency with respect to minute amplitude of the braking force, small amplitude of the wheel speed decreases abruptly in a state near the peak μ Based on the resonance gain G _d
There is regarded as the state just before the peak μ when became equal to or less than the reference gain G _s, control is performed to reduce the braking force.

[0004] Further, a friction state calculating device for calculating a friction state between a wheel and a road surface using the above principle has been proposed (Japanese Patent Application No. 8-157275). According to this friction state calculation unit, reference gain G _s of the resonance gain G _d
Is determined as the braking force corresponding to the peak μ, and the calculation for obtaining the maximum friction coefficient is performed based on the braking force.

Further, according to the friction state calculating device,
If the brake torque is small excitation, to excite vibration of the brake torque T _b appears in the wheel speed small vibrations, by utilizing the fact that the delay constant phase angle theta _d, to simplify the operation of the small amplitude of the wheel speed The method is shown.

That is, in the conventional friction state calculation device,
By a simple method of calculating a correlation from a single sine wave of constant phase angle theta _d in the time series data of the wheel speed alone, the method for obtaining a wheel speed small amplitude of the resonance frequency component is shown. According to this prior art, an attention is paid only to the vibration component delayed relative excitation frequency of the brake torque by a phase angle theta _d, it is possible to extract only a vibration component having a correlation with the excitation of the braking torque Therefore, it is said that the apparatus is hardly affected by vibration noise from the road surface and the like, and the calculation accuracy of the resonance gain is improved.

Note that the phase angle θ _d is the tire spring constant K
And the slope of the coefficient of friction μ against the slip ratio (hereinafter “μ
Gradient α ”).
In the prior art, the delay phase angle theta _d, is treated as a constant value of about 180 degrees ([pi).

Here, the μ gradient α is variously changed (α =
Bode diagram of the tire resonance characteristics when three types (5, 10, 100) are made (based on the transfer function of equation (14) described later)
Are shown in FIGS. 6A and 6B.

FIG. 6A shows a tire resonance model (FIG. 4).
), And has a maximum gain at any resonance frequency (about 40 Hz) of a vibration system (at the time of tire grip) composed of the vehicle body, the wheels, and the road surface for any μ gradient α. You can see that. Further, the value of the gain at this resonance frequency decreases as the μ gradient α decreases.

Meanwhile, FIG. 6 (b), shows the frequency characteristic of the phase angle theta _d, the resonant frequency in the tire grip (approximately 40 Hz), the phase angle theta _d for any α brake The delay angle is always -180 degrees with respect to the torque, that is, the phase is reversed. Therefore, as in the prior art, the time-series data of the wheel speed, by calculating a correlation from only a single sine wave of phase angle theta _d antiphase to the excitation vibrations of the brake torque, the calculation accuracy can be improved I understand. Incidentally, the rate of change of phase angle theta _d of the resonance frequency near, it can be seen that as α is small small.

In addition to the above-described prior art other than the control of the braking force based on the resonance gain, US Pat.
No. 538 (Dec. 27, 1988) discloses a technique for estimating a friction coefficient μ between a road surface and a wheel from a wheel cylinder pressure and a wheel speed, and controlling a braking force acting on the wheel based on the μ value. It has been disclosed. In this technology, three parameters (p ₂ , p ₁ , p ₁ , p ₂ , p ₁ , p ₂ , p ₁ ,
c), and based on the parameter p ₂ ,
The μ gradient is obtained, and the μ value is calculated from the μ gradient. According to this technique, it is possible to accurately determine the friction coefficient μ and μ gradient for each road surface.

[0012]

However, the above-mentioned conventional ABS device and friction state detecting device have the following problems.

[0013] That is, Taiyabane constant K is changed when the tire air pressure or the like is changed, therefore the phase angle theta _d is changed can be easily predicted, in the above prior art, deals with the phase angle theta _d as a constant value Therefore, it is impossible to cope with the change in the tire pressure.

FIG. 7 is a Bode diagram of the tire resonance characteristics when the tire spring constant K is reduced by 20%.
(A) and FIG. 7 (b). In these figures, the frequency characteristics are obtained under the same conditions as those in FIGS. 6A and 6B except for the tire spring constant K.

As shown in FIG. 7A, the resonance frequency at the time of tire grip when the tire spring constant K is reduced by 20% is shifted to a value smaller than the resonance frequency (about 40 Hz) in FIG. 6A. I do. Further, as shown in FIG. 7B, when the tire spring constant K is reduced by 20%, the phase angle θ _d at the resonance frequency (about 40 Hz) at the time of tire grip greatly deviates from the antiphase of −180 degrees. , -200
Degrees to -300 degrees (depending on α). Therefore, in the related art for extracting only the vibration component having the opposite phase, the component having little correlation with the excitation of the braking force is focused on, so that it is affected by the noise, and the calculation accuracy of the resonance gain is reduced. There is a problem.

In general, when an unknown parameter is identified by applying a system identification technique, a calculation amount proportional to the square of the number of parameters to be identified is required, and the identification accuracy also deteriorates as the number of parameters increases. There is. Therefore, in the prior art of the above-mentioned U.S. Patent, since three parameters must be estimated at the same time, there is a problem that the amount of calculation increases and the calculation accuracy deteriorates.

SUMMARY OF THE INVENTION The present invention has been made in view of the above-described circumstances, and has as its object to provide a resonance gain calculating device capable of always calculating a resonance gain with high accuracy irrespective of a variation in a tire spring constant. .

[0018]

[MEANS FOR SOLVING THE PROBLEMS] To achieve the above object
The invention according to claim 1 comprises a vehicle body, wheels, and a road surface.
Micro-excitation of brake torque at the resonance frequency of the vibration system
Excitation means and brake torque for detecting brake torque
Wheel detection means for detecting wheel speed signals
And a step, which is detected based on a resonance model of the vibration system.
From the brake torque and the detected wheel speed signal,
Slip speed of friction coefficient μ which is an unknown constant of resonance model
Estimating means for estimating gradient and tire spring constant with respect to tire
And the slip of the friction coefficient μ estimated by the estimation means.
Based on the slope versus the tire spring constant
Micro vibration of brake torque excited by recording means
To calculate the phase from the signal to the minute vibration of the wheel speed signal
Calculating means for calculating the phase from the detected wheel speed
Car having a phase calculated by a step and the resonance frequency
An amplitude detecting means for detecting an amplitude value of the wheel speed minute vibration;
The amplitude value of the minute vibration of the brake torque and the amplitude detection
Of amplitude of wheel speed minute vibration detected by means
And a gain calculating means for calculating
is there. (Principle of the present invention) As shown in FIG.
Vehicle speed v (ω in angular velocity conversion) _v)
Vibration phenomena at the wheels when the vehicle is on, ie the body, wheels and road surface
The vibration phenomenon of the vibration system composed of
Referring to the resonance model shown in FIG.
To consider.

In the resonance model shown in FIG. 4, the braking force acts on the road surface via the surface of the tread 115 of the tire in contact with the road surface, but this braking force is actually applied to the vehicle body as a reaction (braking force) from the road surface. Therefore, the equivalent model 117 of the vehicle body weight in terms of the rotation axis is connected to the wheel 113 via the frictional element 116 (road surface μ) between the tread of the tire and the road surface. This is the same as the large inertia under the wheels, that is, the weight of the vehicle body can be simulated by the mass on the side opposite to the wheels, as in the chassis dynamo device.

FIGS. 3 and 4 show a wheel 11 including a tire rim.
3 of inertia J _w, K the spring constant of the spring element 114 between the rim and the tread 115, the inertia of J _t of the tread 115, the friction coefficient of the frictional element 116 between the tread 115 and the road surface mu, the vehicle body the inertia of the equivalent model 117 of the 112 weight of the rotating shaft conversion When J _V, of the whole system characteristic follows
(1) to (3). In the following, the first derivative d / dt relating to time is represented by “′”, and the second derivative d ² / dt ² relating to time is represented by “”.

[0021] _{J W θ w "= -T b} + K (θ t -θ w) (1) J t θ t" = -K (θ t -θ w) + μWR (2) J v ω v '= -μWR (3) Here, ω = θ _w ′ (4) J _v = R ² W (5) ω _v = v / r (6), θ _w is the rotation angle of the wheel 113, and θ _w ”is the wheel 11
3 of angular acceleration, omega is the angular velocity of the wheel 113, i.e. the wheel speed, the rotation angle of theta _t tread 115, theta _t "
Is the rotational angular acceleration of the tread 115, ω _v is the rotational angular velocity in terms of the rotational axis of the vehicle equivalent model 117, and T _b is the wheel 113.
, W is the weight of the vehicle body, and R is the wheel radius. Incidentally, the braking torque T _b is performed actually under the control of the braking pressure P _b.

Here, when the vehicle is running at a certain speed, when the brake is applied, a slip occurs between the wheel and the road surface. The friction coefficient μ between the wheel and the road surface is determined by a slip speed Δω. (= Ω _v −ω) as shown in FIG.

When the tire is gripping while traveling on a road surface having such a μ-Δω characteristic, it is considered that the tread 115 and the vehicle equivalent model 117 shown in FIG. Inertia and tread 115
And the inertia of the wheel 13 resonates with the inertia of the wheel 13, and the resonance frequency ω∞ of the wheel resonance system at this time is

[0024]

(Equation 1) Becomes This state corresponds to the slip speed region A1 before reaching the peak μ in FIG. 8, and the μ gradient of the friction coefficient μ with respect to the slip speed Δω at this time is a positive value.

Conversely, when the friction coefficient μ of the tire approaches the peak μ, the friction coefficient μ of the tire surface hardly changes with respect to the slip speed Δω, and the component accompanying the inertial vibration of the tread 115 is equivalent to the vehicle equivalent. It does not affect the model 117. That is, the tread 115 is equivalent to the vehicle equivalent model 1
17 is separated, and the tread 115 and the wheel 113 resonate. The resonance frequency ω 共振 ′ of the wheel resonance system at this time is

[0026]

(Equation 2) Becomes

In FIG. 8, this state corresponds to the area A2 of the slip speed near the peak μ, and generally, when the point of the peak μ is reached, the state immediately transits to the area A3 and the tire is locked. On the other hand, the gain of the wheel speed at the resonance frequency also sharply decreases immediately before the peak μ. In the region A2 in FIG. 8, the μ gradient has a value near 0, and in the region A3, the μ gradient has a negative value.

[0028] The magnitude of each inertia is _{J t <J w <J v} (9), than this, become ω∞ <ω∞ '(10). That is, when the tire is locked, the resonance frequency of the wheel resonance system shifts to the high frequency side. Also,
This change in the resonance frequency occurs sharply near the peak μ.

The simplified model, the change in the peak of the gain of the resonance frequency and the wheel speed of approaching the peak μ state even when ignoring the inertia J _t of sufficiently small tread 115 wheel resonant system relative to the wheel inertia J _w Occur and a similar analysis is possible.

Therefore, in the present invention, the following resonance gain _Gd is introduced as a physical quantity reflecting the above-described change in the resonance frequency. In the following, the resonance gain G
The _d, but represented using a brake torque T _b, may be alternatively expressed with braking pressure P _b in the substantially proportional relationship with the braking torque T _b.

First, the excitation means according to the present invention has a resonance frequency ω
When the brake torque (or the brake pressure) is slightly excited at ∞, the wheel speed ω also slightly vibrates around the average wheel speed at the resonance frequency ω∞. At this time, the resonance gain G
the _d, the ratio of the wheel speed ω to the brake torque T _b (ω
/ T _b ), G _d = ((ω / T _b ) | s = jω∞) (11) Here, j is an imaginary number, and s is a Laplace operator. Incidentally, at the resonant frequency Omega∞, a small amplitude value of the braking torque T _b T _v, if the small amplitude value of the wheel speed omega and the omega _v, resonance gain G _{d is, G d = ω v / T} v (12 ).

The resonance gain G _d is related to the slope α of the friction coefficient μ between the tire and the road surface with respect to the slip speed,
(1) Based on the formula - (3) transfer relationship derived from formula, G _d = jA + αB represented as (j is an imaginary unit) (13). However,

[0033]

(Equation 3) It is.

[0034] (13) based on the equation, the phase theta _d of the resonant gain, i.e. the phase theta _d from small vibrations of the brake torque is excited by the excitation means to the small vibration of the wheel speed signals for calculating according to the following formula Can be.

Θ _d = tan ⁻¹ (A / αB) (16) In this equation (16), the gradient α of the friction coefficient μ with respect to the slip speed, which is an internal parameter of the resonance gain G _d , and the tire spring constant are unknown constants. Contains K. Therefore,
The estimating means of the present invention uses the detected brake torque _Tb and the detected wheel speed signal ω based on the resonance model of the vibration system of FIG. The gradient α for the speed and the tire spring constant K are estimated.

Thus, the phase calculating means can be used as the estimating means.
Slope of friction coefficient μ estimated from slip with respect to slip speed
And (14) to (16) based on the tire spring constant
Phase θ _dCan be calculated.

Then, the amplitude detecting means calculates the amplitude ω of the wheel speed minute vibration having the phase θ _d and the resonance frequency ω∞ calculated by the phase calculating means from the detected wheel speed ω.
_v is detected and the gain calculating means, the resonance gain G which is the ratio of the amplitude value T _v of the micro vibration of the brake torque, the amplitude value omega _v of the wheel speed small vibration detected by the amplitude detecting means
Calculate _d (Equation (12)).

As a method for simultaneously estimating the tire spring constant K and the .mu. Gradient .alpha., Which are unknown constants, in the invention of claim 2, the estimating means of claim 1 comprises a step of: The communication relationship of
Expressed such that a linear relationship is established with respect to the physical quantity related to the slope and the tire spring constant of the friction coefficient μ, which is an unknown constant, and is expressed by the linear relationship,
By applying an online system identification method to each data obtained by applying discretized data of the detected brake torque and discretized data of the detected wheel speed signal to the discretized transmission relationship. , Friction coefficient μ
And the tire spring constant is calculated with respect to the gradient of the slip speed.

A more specific calculation method according to the second aspect of the present invention is as follows. (Calculation method according to the second aspect of the invention) When the transmission relationship from the brake torque _Tb to the wheel speed ω is represented based on the resonance models of FIGS. 3 and 4, the following transfer function is obtained. In the following equation model, by ignoring the inertia J _t of sufficiently small tread 115 as compared with the wheel inertia J _w,
For simplicity.

[0040]

(Equation 4) Here, α is a μ gradient, and s is a Laplace operator.

The transfer function of equation (17) can be modified as follows.

[0042]

(Equation 5) Therefore, in order to apply online system identification methods such as the least squares method and the auxiliary variable method, the transfer relation of Eq.
Physical quantity related to slope α and tire spring constant K for slip speed of friction coefficient μ

[0043]

(Equation 6) Is represented by a linear relationship such that φ ^T · θ = y (20), where “ ^T ” is the permutation of the matrix. However,

[0044]

(Equation 7) It is.

Here, φ and y are continuous quantities, but in order to apply the system identification method, it is necessary to discretize equations (20) to (22). That is, the Laplace operator s is represented by an operator of the z-transform using a discretization method such as bilinear transformation or backward difference, and the brake torque _Tb and the wheel speed signal ω are discretized to obtain the time. Expressed by series data.

In this way, the transmission relation obtained by discretizing the φ and the y is φ [k] ^T · θ = y [k] (k is a sample number: k = 1, 2, 3,...) (23) The discretized data of the detected brake torque T _b [k]
(k = 1,2,3, ...) and the discretized data ω [k] (k = 1,2,3, ...) of the detected wheel speed signal On the other hand, by applying an online system identification method, the above θ is estimated, and the friction coefficient μ is calculated from each element of the θ.
The slope α and the tire spring constant K with respect to the slip speed are calculated.

Since the present invention focuses on the behavior of the frequency component in the band including the resonance frequency ω∞, in order to maintain the identification accuracy with high accuracy, the detected brake torque and the wheel speed signal are used to determine the accuracy. It is desirable to cut out frequency components outside the band in advance.

Here, the least squares method can be applied as one of the system identification methods. That is,

[0049]

(Equation 8) And when

[0050]

(Equation 9) From the recurrence formula of

[0051]

(Equation 10) Is calculated. Here, λ is a forgetting constant.

[0052]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment in which the resonance gain calculation device of the present invention is applied to an ABS device of a vehicle will be described below in detail with reference to the drawings. (First Embodiment) FIG. 1 is a block diagram showing a configuration of an ABS apparatus according to a first embodiment of the present invention.

As shown in FIG. 1, A of the first embodiment
BS device, A to each wheel as the resonant gain calculation unit 10 for calculating a resonant gain G _d for each wheel, computed resonance gain G _d is coincident or substantially coincident with the reference gain G _s
An ABS control unit 18 that calculates and outputs a BS operation amount signal (an average brake pressure command _Pr ), and a control valve 52 that controls a brake pressure (wheel cylinder pressure) for each wheel based on the ABS operation amount signal. A tire pressure diagnostic unit 24 for diagnosing the tire pressure of each wheel based on the calculated tire spring constant K of each wheel, and a diagnostic result output unit 26 for outputting the diagnostic result of the tire pressure so that the driver can recognize it. , Consisting of

Here, the ABS control unit 18 sets the resonance gain
G_dAnd reference gain G_sAnd the difference (G _d-G_s) To zero
A so-called PI controller or PID for matching or nearly matching
It can be realized by a controller. Other than this
For example, robust control such as H∞ control and two-degree-of-freedom control
System, neural computer, fuzzy control system, application
It is also possible to configure using control or the like.

Further, the ABS control unit 18 generates a small excitation command _Pv for minutely exciting the brake pressure at the resonance frequency ω∞ of the vibration system composed of the vehicle body, the wheels, and the road surface or at an excitation frequency near the frequency. It is generated and output to the control valve 52.

Further, the tire pressure diagnostic unit 24 classifies the calculated tire spring constant K into one of a plurality of groups determined in advance according to a value of a possible range of the tire spring constant, or It can be configured by software processing. As this specific processing, for example, the calculated tire spring constant K is set to a reference value K _s.
When the above, classified into groups of "tire pressure normal", is smaller than the reference value K _s are classified into groups of "tire pressure abnormality", diagnose this classification result result output section 2
Transmit to 6.

The diagnosis result output unit 26 can be constituted as a display means such as a liquid crystal display for displaying the transmitted diagnosis result, or as an output means for synthesizing and outputting the diagnosis result.

[0058] Further, the resonant gain calculation unit 10, a brake torque sensor 14 for detecting a brake torque T _b acting on the wheel, the wheel speed sensor 16 for detecting a wheel speed signal omega, the detected braking torque T _b A μ gradient / tire spring constant calculating unit 12 that calculates a μ gradient α and a tire spring constant K based on the detected wheel speed signal ω,
a phase calculating unit 15 from the μ gradient α and Taiyabane constant K for calculating the delay phase angle theta _d from small excitation frequency of the brake torque to the wheel speed signal, the detected braking torque T _b and the detected wheel speed signal ω based on the bets, in consideration of the delay of the operation phase theta _d, a gain calculation unit 17 for calculating a resonant gain G _d, and a.

Here, as a method of detecting the brake torque by the brake torque detector 14, for example, in addition to a method of directly detecting the brake torque, the brake torque is estimated from a wheel cylinder pressure detected by a pressure sensor or the like in consideration of an operation delay. There is a method of estimating from an ABS operation amount signal as a brake pressure command in consideration of an operation delay, and the like.

Next, the μ gradient / tire spring constant calculator 12
Will be described with reference to FIG.

[0061] As shown in FIG. 2, mu gradient-Taiyabane constant calculating unit 12, the analog signal of the detected braking torque T _b, discretized at every predetermined sample time τ to cut high frequency components AD the transducer 27, the resonant frequency ω∞ from the time-series data of the discretized braking torque T _b
And a band-pass filter 29 that passes only the frequency component of the band including the detected wheel speed signal ω.
An AD converter 28 that discretizes each wheel and a band-pass filter 30 that passes only frequency components of a band including the resonance frequency ω か ら from the time-series data of the discretized wheel speed are provided.

The band-pass filters 29 and 30 are digital filters designed so that their bandwidths fall within the range of fluctuation of the resonance frequency due to the expected fluctuation of the tire spring constant K. It is connected to the operation unit 31. These bandpass filters 2
According to 9 and 30, attention is paid only to the frequency components of the bandwidth including the resonance frequency, so that the influence of other frequency components is reduced, and a decrease in identification accuracy is prevented.

The parameter calculation unit 31 obtains φ by discretizing φ in the equation (21) for each sample time τ.
[K], y [k] obtained by discretizing y in equation (22) for each sample time τ is converted to bandpass filter 2
9 based on the output T _b [k] (k = 1, 2, 3,...) And the output ω [k] (k = 1, 2, 3,...) Of the bandpass filter 30. Circuit.

As a specific operation method, when discretizing by the so-called bilinear z-transform, the Laplace operator s

[0065]

[Equation 11] And substitute it into Eqs. (21) and (22).
_b, omega, and each _{T b → T b [k]} ω → ω using equation obtained by converting each time-series data as [k]. However, z ^-1 operator to mean a delay of one sample time _{^{τ, z -1 T b [k}} ] = T b [k + 1] z -1 ω [k] = ω replaced by [k + 1]. Note that each element of φ includes not only s but also s
^{Since 2} and s ³ are included, time-series data delayed by 2 samples and 3 samples is also used.

The output end of the parameter operation unit 31 receives the calculated time series data of φ [k] and y [k],
By using online least squares method for each data obtained by substituting into equation (23), (2
A θ estimating unit 32 for calculating θ in Expression 3) is connected. The θ estimating unit 32 calculates L [k], P by equations (24) and (25).
It is configured as a circuit that calculates [k] and further calculates an estimated value of θ from the recurrence formula of Expression (26). As described above, the μ gradient α and the tire spring constant K are estimated based on the resonance model that sensitively reflects the slip state between the tire and the road surface, and the number of identification parameters is suppressed to two, so that the calculation accuracy is maintained with high accuracy. And the calculation time can be reduced.

The output end of the θ estimating section 32 is connected to an output section 33 for calculating the μ gradient α and the tire spring constant K from the estimated value of θ. The output unit 33 includes a divider (not shown) for calculating a tire spring constant K by dividing 1 by a first element (1 / K) of θ represented by Expression (19), and a second element (1 / Α), and a divider (not shown) that calculates the μ gradient α by dividing 1 by 1.

Although not shown in FIGS. 1 and 2, the resonance gain calculating device 10 described above is provided for each wheel.

By the way, the control of the brake pressure by the control valve 52 can be realized by the brake section 22 provided around the control valve 52. Here, a detailed configuration of the brake unit 22 is shown in FIG.

As shown in FIG. 5, the brake section 22 includes a master cylinder 48, a wheel cylinder 56, a reservoir 58, and an oil pump 60 in addition to the control valve 52. Note that a control valve 52 is also provided for each wheel. The brake pedal 46 is connected to a pressure-intensifying valve 50 of a control valve 52 via a master cylinder 48 that increases the pressure according to the depression force of the brake pedal 46. The control valve 52 is connected to a reservoir 58 as a low pressure source via a pressure reducing valve 54. Further, a wheel cylinder 56 for applying the brake pressure supplied by the control valve to the brake disc is connected to the control valve 52. This control valve 5
2 supplies the brake pressure P _d by pedaling force of the driver, controls the opening and closing of the pressure increasing valve 50 and pressure reducing valve 54 based on the valve operation command input (P _r + P _v).

The control valve 52 is connected to the pressure increasing valve 5
0 is controlled to open only the wheel cylinder 5
The hydraulic pressure (wheel cylinder pressure) of the master cylinder 48 (master cylinder pressure) is proportional to the pressure obtained when the driver depresses the brake pedal 46.
To rise. Conversely, when the pressure is controlled so as to open only the pressure reducing valve 54, the wheel cylinder pressure decreases to the pressure of the reservoir 58 (reservoir pressure), which is approximately atmospheric pressure. Further, when both valves are controlled to be closed, the wheel cylinder pressure is maintained.

The braking force applied to the brake disk by the wheel cylinder 56 includes a pressure increasing time during which the high hydraulic pressure of the master cylinder 48 is supplied, a pressure reducing time during which the low hydraulic pressure of the reservoir 58 is supplied, and a holding pressure during which the supplied hydraulic pressure is maintained. It is determined from the time ratio and the master cylinder pressure and the reservoir pressure detected by a pressure sensor or the like.

Accordingly, a desired braking force can be realized by controlling the increasing / decreasing time of the control valve 52 according to the master cylinder pressure. The micro-excitation of the brake pressure is controlled by the control valve 5 which realizes the average braking force
This can be achieved by performing the pressure increase / decrease control at a cycle corresponding to the resonance frequency simultaneously with the pressure increase / decrease control of (2).

FIG. 9 shows the specific contents of the control.
As described above, a half period of the period of the minute excitation (for example, 24 [ms])
Each mode of pressure increase and pressure reduction is switched every T / 2,
The pressure increase / decrease command to the valve increases from the moment the mode is switched.
Pressure time t_i, Decompression time t_rPressure for each time
・ Output the pressure reduction command and output the hold command for the remaining time.
You. The average braking force is increased according to the master cylinder pressure.
Time t_iAnd decompression time t _rIs determined by the ratio of
Pressure increase / decrease mode every half cycle T / 2 corresponding to resonance frequency
Switch, the micro-vibration around the average braking force
Motion is applied.

The phase calculator 15 can be configured as a calculator for calculating the equation (16). Although not shown,
For example, using a tire spring constant K calculated by the μ gradient / tire spring constant calculation unit 12, a calculator for calculating the parameters A and B according to equations (14) and (15), and a calculated μ
A multiplier that multiplies the gradient α by the parameter B, a divider that divides the parameter A by the multiplication result αB, and a division result A /
and a table that converts α into an input value θ and outputs the result as tan ⁻¹ θ.

Next, a detailed configuration of the gain calculator 17 will be described with reference to FIG. Note that the gain calculation unit 17
Is configured to detect a predetermined vibration component. For example, when a detection signal is read into the computer by a so-called AD converter or the like, if the operation sampling time of the computer is set to 1 [ms], a half cycle is obtained. 12 sample points,
The aforementioned vibration component of 41.7 [Hz] is detected.

As shown in FIG. 10, the wheel speed sensor 16
Input terminal (shown in the figure)
There is no BPF for removing DC and noise components
(So-called band pass filter) 130
The number detection unit 134 is connected. This correlation coefficient detector
134 is the amplitude ω of the resonance frequency component ω∞ of the wheel speed ω _v
Is calculated.

[0078] Further, to the input terminal of braking torque T _b detected by the braking torque detecting unit 14 is input (not shown), the DC component and BP for noise component removal
The correlation coefficient detection unit 136 is connected via F132. The correlation coefficient detection unit 136 determines whether the braking torque T _b
The amplitude T _v of the resonance frequency ω∞ component is calculated.

Further, correlation coefficient detecting sections 134 and 136
The output end, the divider 138 is connected to dividing the output signal omega _v output signal T _v. Note that this divider 13
The division result of 8 is the resonance gain G _d = ω _v / P _v .

Here, assuming that the time length of the time series data is T ₁ , the minimum frequency of the obtained frequency spectrum is 1 / T
₁ [Hz], and an integer multiple of this frequency component can be obtained. Assuming that the sampling time of the control system is 1 [ms], 48
Using the data of the sample points, the minimum frequency is about 20.
8 [Hz], and the second component is about 41.7 [Hz]. The correlation coefficient detection units 134 and 136 can be configured as shown in FIG. 11 to calculate the power of the 41.7 [Hz] component.

As shown in FIG. 11, the present correlation coefficient detecting section performs the processing on the time-series data {x ₁ ,
x ₂ ,. . . . . , X _n } are delayed by one sample and output, and the sample data x ₁ , x ₂ ,. . . , Factor x ₄₈ cs _1, c
s ₂ ,. . . , Cs ₄₈ , respectively.

The delay circuit 150 has a configuration in which a plurality of unit delay elements z ^-1 for delaying time series data by one sample each are connected in series. This unit delay element z
The number of ^-1 is set to match the number of samples 48 of the time-series data. An output signal line is provided for sending each sample data to each coefficient multiplier of the coefficient multiplication unit 160 before the time series data is input to each unit delay element z ^-1 .

The output terminal of each coefficient multiplier of the coefficient multiplying section 160 is connected to an adder 162, and this adder adds all the multiplication results of each coefficient multiplier and outputs the result.

At this time, the output y is

[0085]

(Equation 12) Becomes Here, the coefficient cs _i is,

[0086]

(Equation 13) It is. Here, θ _c is the phase of the brake torque with respect to the small excitation signal _Pv (for example, the phase based on the pressure increase start time of the valve operation command in FIG. 9), and in the case of Expression (29), this phase θ _c is represented by a discrete value every sampling period T _1. Here, the correlation coefficient detection unit 136 sets θ _c = 0 assuming that there is no delay of the actual brake torque with respect to the minute excitation signal _Pv, and sets the correlation coefficient calculation unit 134
So it is set to a phase theta _d calculated by the phase calculating section _{15 (θ c = θ d)} . Thus, in the present embodiment,
Since the phase of the minute vibration of the brake torque and the phase of the minute vibration of the wheel speed with respect to the minute excitation signal _Pv are known,
The correlation can be calculated by finding only the Fourier coefficient for a single component having a frequency of 41.7 [Hz] as in the equation (28), thereby simplifying the device configuration.

Next, the operation of the first embodiment will be described.

When the predetermined condition is satisfied, a micro-excitation command _Pv is applied to the operation command to the control valve 52, and the brake pressure is micro-excited near the resonance frequency ω∞. Note that the predetermined condition includes a normal ABS start condition, for example, a condition in which the driver depresses the brake pedal and the wheel deceleration exceeds a certain value.

[0089] When the brake pressure is small excited, the braking torque is minute vibrations acting on the wheel, the wheel speed ω is vibrated by the vibration component of the braking torque T _b. Here, the road μ gradient calculating unit 10 of FIG. 1, on the basis of the transfer relationship of equation (17) from the braking torque T _b obtained by the resonance model of Figure 4 to the wheel speed omega, the detected braking torque T _{From b} and the wheel speed signal ω, the unknown constant μ
The gradient α and the tire spring constant K are simultaneously estimated and calculated by the online least squares method.

The phase calculator 15 calculates the estimated μ gradient α
And based on the Taiyabane constant K, to calculate the phase difference theta _d from small excitation frequency of the brake torque to the small vibration of the wheel speed.

The gain calculator 17 calculates the correlation coefficient between the minute excitation signal _Pv and the single sine wave having a phase difference of 0 from the time series data of the brake torque, and calculates the minute vibration amplitude T of the brake torque. _v and the phase difference θ _d from the minute excitation signal P _v from the time series data of the wheel speed.
By calculating the correlation coefficient between the single sine wave to calculate a small vibration amplitude omega _v of the wheel speeds of the further calculates the resonance gain G _d is these ratios. At this time, since only the vibration component of the wheel speed that is correlated with the excitation of the brake pressure is extracted,
It is hardly affected by the road surface vibration noise, and it is possible to detect only the resonance characteristics of the vibration system including the vehicle body, the wheels, and the road surface.

[0092] The ABS control unit 18, the calculated resonance gain G _d is the ABS operation amount signal (mean braking pressure command P _r) such as to match or substantially match the set G _s as a reference value for calculating the output . In the control valve 52, the command P
Control the pressure increase / decrease time of the valve so as to realize the average brake pressure corresponding to _r .

[0093] Here, the reference gain G _s, the friction coefficient μ is set to the value of μ gradient at A2 region to be a substantially peak value (positive value close to zero) in FIG. 8, control following the peak μ is It becomes possible. In the vicinity of the peak μ, the fact that the value of the μ gradient becomes close to 0 is invariant regardless of the road surface condition.
A stable anti-lock brake operation is always possible.

In the tire pressure diagnostic section 24, the tire spring constant K calculated by the road μ gradient calculating device 10 is calculated.
Diagnosis of tire pressure is performed based on Then, the diagnosis result output unit 26 outputs the diagnosis result of the tire pressure in a format that can be recognized by the driver. As a result, the driver can determine whether or not to inflate the tire, so that safe driving is possible. It should be noted that the tire pressure diagnostic unit 24 may be configured to diagnose whether the tire pressure is suitable for traveling on a highway.

As described above, in this embodiment, even if the phase difference θ _d from the minute excitation vibration of the brake torque to the minute vibration of the wheel speed deviates from −180 degrees due to the fluctuation of the tire spring constant K (FIGS. 6 and 7). ), The phase is estimated from the brake torque and the wheel speed, and the correlation between the single sine wave having the estimated phase and the wheel speed signal is calculated. Therefore, the phase θ _d is fixed at −180 degrees. compared with the conventional method was calculated the resonance frequency component of the wheel speed, the detection sensitivity can be obtained satisfactorily maintaining the as accurate resonance gain G _d. Accordingly, a highly accurate anti-lock brake operation can be performed regardless of the fluctuation of the tire pressure. (Second Embodiment) In a second embodiment, an example is shown in which the correlation coefficient detection units 134 and 136 in FIG. 10 are configured more simply. In addition, about the structure similar to 1st Embodiment, the same code | symbol is attached | subjected and detailed description is abbreviate | omitted.

FIG. 13 has a period of about 41.7 [Hz].
The graph of the single sine wave of Formula (29) is shown. As shown in FIG. 13, a sample idx of the input signal x (discrete index of x)
Increases monotonically from 1 to 24 every 1 msec for each sample in synchronization with the minute excitation vibration _Pv of the brake torque, and is reset to 1 after 24. The phase θ _c of the amplitude CS _n of the single sine wave with respect to idx is set to 0 when the input signal is a brake torque sample signal. Then, if the input signal is the wheel speed of the sample signal is a phase theta _d calculated by the phase calculating section 15. That is, a phase obtained by shifting the phase of the amplitude CS _n by θ _d is used.

Assuming that the time point of the first 48 samples is the time point j, y ^j is given by equation (28).

[0098]

[Equation 14] Becomes Here, cs ^j _i and x ^j _i are obtained by assigning numbers from 1 to 48 from the past values to the data of the past 48 points at the time point j. Here, the value y ^{J + 1} at the next sample time point j ^{+ 1} is

[0099]

(Equation 15) Becomes Obviously, cs ^{j + 1} _i = cs ^j _{i + 1} , x ^{j + 1} _i = x ^j _{i + 1} (32), and due to the periodicity of cs ^j _i , y ^{j + 1} = y ^j + Cs ^{j + 1} ₄₈ (x ^{j + 1} ₄₈ −x ^j ₁ ) (33) is obtained.

Cs ^{j + 1} ₄₈ is CS _n itself at the time point j + 1, and x ^{j + 1} ₄₈ is the latest value of x at the time point j + 1, x ^j
_{Since 1} is a value 48 samples before at the time of j + 1, the correlation coefficient detecting unit can be configured as shown in FIG.

As shown in FIG. 12, the correlation coefficient detecting section according to the second embodiment includes a delay element 170 for ^obtaining a signal value x ⁻⁴⁸ which is 48 samples before the input signal x, Means 172 for ^calculating the difference between x and ^-48, and id
computing means 174 for computing CS _n for x, and output y
Delay element 178 for finding signal value y ^-1 one sample before
And an adder 176 for calculating the sum of the outputs of the delay elements 178, and can calculate the result of equation (33). Further, the correlation coefficient detection unit 1
For 34, the calculating means 174 has a function of shifting the CS _n in accordance with the input phase theta _d.

According to this configuration, the amount of calculation is only two additions and one multiplication per sample, and the processing can be further simplified and speeded up.

The embodiments of the present invention have been described above. However, the present invention is not limited to the above-described examples, and can be arbitrarily and suitably changed without departing from the gist of the present invention.

For example, in the θ estimator 32 of FIG. 2, the online least squares method is used, but an auxiliary variable method, which is another system identification method, may be used. Further, in the parameter operation unit 31, as the discretization method of φ and y, the bilinear z-transform with a small aliasing error is used.
Any method that can convert φ, y (Equations (21), (22)) in the region into the z region can be applied, and for example, a method such as a backward difference can also be used.

As a method of minutely exciting the brake pressure,
The control is performed by switching the pressure increase / decrease mode of the control valve. However, the present invention is not limited to this.
It is also possible to use a piezoelectric element that directly applies the excitation brake pressure according to _v to the brake disk.

Further, the minute amplitude T of the brake torque
_v is between the master cylinder pressure detected by the pressure sensor or the like, increase of the control valve pressure time t _i and the pressure reducing time t _r (FIG. 9
See also).

[0107]

As described above, according to the present invention,
Based on the gradient of the estimated friction coefficient μ with respect to the slip speed and the tire spring constant, the phase from the minute vibration of the excited brake torque to the minute vibration of the wheel speed signal is calculated and has the phase and the resonance frequency of the vibration system. Since the amplitude of the wheel speed minute vibration is detected and the resonance gain is calculated based on the amplitude value, the phase difference from the minute excitation vibration of the brake torque to the minute wheel speed vibration is -180 degrees due to the fluctuation of the tire spring constant. Even if it deviates, the excellent effect that the resonance gain corresponding to the phase fluctuation can always be calculated with high accuracy can be obtained.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a configuration of an ABS device to which a road μ gradient calculating device according to a first embodiment of the present invention is applied.

FIG. 2 is a block diagram showing a detailed configuration of a μ gradient / tire spring constant calculation unit which is a main component of the road μ gradient calculation device according to the first embodiment of the present invention.

FIG. 3 is a diagram showing a dynamic model of a vehicle.

FIG. 4 is a diagram showing a resonance model equivalent to a vibration system composed of wheels, a vehicle body, and a road surface.

FIG. 5 is a block diagram illustrating a hardware configuration of a brake unit.

FIG. 6 is a Bode diagram of a tire resonance characteristic, and FIG.
The change in gain with respect to frequency, shows the change in phase angle theta _d for (b) the frequency.

FIG. 7 is a Bode diagram of tire resonance characteristics when the tire spring constant of the tire of FIG. 6 is reduced by 20%;
(A) shows the gain variation of the relative frequency, a change in phase angle theta _d for (b) the frequency.

FIG. 8 is a diagram showing a change characteristic of a friction coefficient μ and a μ gradient α with respect to a slip speed Δω.

FIG. 9 is a diagram showing an operation command to a control valve according to the first embodiment of the present invention.

FIG. 10 is a block diagram illustrating a detailed configuration of a gain calculating unit 17 in FIG. 1;

11 is a block diagram illustrating a detailed configuration of the correlation coefficient detection units 134 and 136 of FIG. 10 according to the first embodiment.

FIG. 12 is a block diagram illustrating a detailed configuration of a correlation coefficient detection unit according to a second embodiment in FIG. 10;

FIG. 13 is a graph of a single sine wave having a period of about 41.7 Hz.

[Explanation of symbols]

 Reference Signs List 10 road surface μ gradient calculating device 12 μ gradient / tire spring constant calculating unit 14 brake torque detecting unit 16 wheel speed sensor 18 ABS control unit 22 brake unit 29 band pass filter 30 band pass filter 31 parameter calculating unit 32 θ calculating unit 52 control valve 134 Correlation coefficient detector 136 Correlation coefficient detector 138 Divider

 ────────────────────────────────────────────────── ─── Continuing on the front page (72) Inventor Katsuhiro Asano 41-cho, Yokomichi, Nagakute-cho, Aichi-gun, Aichi Prefecture Inside Toyota Central Research Laboratory Co., Ltd. 41, Yokomichi, Toyota Central Research Laboratory Co., Ltd. (72) Inventor Hiroyuki Yamaguchi 41, Ochicho, Yoji, Nagakute-cho, Aichi-gun, Aichi Prefecture 1 Toyota Toyota Central Research Laboratory Co., Ltd.

Claims (2)

[Claims] 1. Exciting means for minutely exciting a brake torque at a resonance frequency of a vibration system composed of a vehicle body, wheels, and a road surface, brake torque detecting means for detecting a brake torque, and wheel speed for detecting a wheel speed signal. Detecting means, based on a resonance model of the vibration system, estimating a gradient and a tire spring constant of a friction coefficient μ, which is an unknown constant of the resonance model, with respect to a slip speed from a detected brake torque and a detected wheel speed signal. Estimating means, based on the slope of the friction coefficient μ estimated by the estimating means with respect to the slip speed and the tire spring constant, the phase from the minute vibration of the brake torque excited by the exciting means to the minute vibration of the wheel speed signal. A phase calculating means for calculating, and a phase calculated by the phase calculating means from the detected wheel speed. Amplitude detecting means for detecting an amplitude value of the wheel speed minute vibration having the resonance frequency; anda ratio between an amplitude value of the brake torque minute vibration and an amplitude value of the wheel speed minute vibration detected by the amplitude detecting means. Gain calculating means for calculating; and a resonance gain calculating device including: 2. The estimating means sets the transmission relationship from the brake torque to the wheel speed based on the resonance model as a linear relationship with a gradient with respect to a slip speed of a friction coefficient μ as an unknown constant and a physical quantity with respect to a tire spring constant. Expressed to be satisfied, expressed by the linear relationship, and discretized transmission relationship, discretized data of the detected brake torque and discretized data of the detected wheel speed signal were applied. 2. The resonance gain calculation device according to claim 1, wherein the gradient of the friction coefficient μ with respect to the slip speed and the tire spring constant are calculated by applying an online system identification method to each data.