JP2010179767A - Road surface friction coefficient estimating device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To always estimate an accurate road surface μ with high responsiveness even in traveling an upward/downward slope. <P>SOLUTION: A road surface μ estimating device 10 computes an actual longitudinal force generated by a tire at each sampling time, as an actual longitudinal force Fmsve, computes a difference value between a present value and a past value of the actual longitudinal force Fmsve as an actual longitudinal force difference value ΔFmsve, computes an ideal longitudinal force generated by the tire based on a tire model including the road surface μ as a parameter at the same timing as the actual longitudinal force Fmsve, as an ideal longitudinal force Fmdve, computes a difference value between a present value and a past value of the ideal longitudinal force Fmdve as an ideal longitudinal force difference value ΔFmdve, and determines a value of the road surface μ by optimization computation to minimize the square sum of a deviation between the actual longitudinal force difference value ΔFmsve and the ideal longitudinal force difference value ΔFmdve at each sampling time at least. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to a road surface friction coefficient estimating device that estimates a road surface friction coefficient based on a tire longitudinal force.

  In recent years, various control techniques for traction control, braking force control, torque distribution control, and the like have been proposed and put into practical use in vehicles. Many of these technologies use the road surface friction coefficient (hereinafter abbreviated as road surface μ) for calculation or correction of the required control amount. In order to reliably execute the control, the accurate road surface μ is estimated. There is a need to.

  For example, in Japanese Patent Application Laid-Open No. 2005-7972 (hereinafter referred to as Patent Document 1), the vehicle body speed is measured, the vehicle body acceleration is calculated, and the difference between the wheel speed and the vehicle body speed of each of the left and right drive wheels is determined. A technique is disclosed in which each slip ratio is calculated, the ratio of each slip ratio to the vehicle body acceleration is taken as a slip ratio, and the road surface μ of each of the left and right drive wheels is estimated from the magnitude of the slope of the slip ratio.

JP 2005-7972 A

  However, as in the technique disclosed in Patent Document 1 described above, in the technique of measuring the vehicle body speed, calculating the slip ratio, and estimating the road surface μ, the necessary parameter is accurately detected and the road surface μ is accurately determined. It is difficult to estimate, and there is a problem that it is difficult to obtain a road surface μ with a desired accuracy because the influence of the road surface gradient is included particularly when traveling on an uphill / downhill. In addition, there is a problem that if the road surface μ is obtained with high accuracy, it cannot be estimated with good response.

  The present invention has been made in view of the above circumstances, and it is an object of the present invention to provide a road surface friction coefficient estimation device that can always accurately estimate a road surface μ with good response even when traveling on an uphill / downhill. .

  In the present invention, the actual longitudinal force generated by the tire at each sampling time is calculated as an actual longitudinal force, and a difference value between the current value of the actual longitudinal force and a past value is calculated as an actual longitudinal force difference value. The ideal longitudinal force generated by the tire is calculated as an ideal longitudinal force by a tire model including a road surface friction coefficient as a parameter at the same timing as the actual longitudinal force, Ideal front / rear force difference value calculating means for calculating a difference value between the current value and the past value as an ideal front / rear force difference value, and at least the actual front / rear force difference value and the ideal front / rear force difference value for each sampling time, Road surface friction coefficient estimating means for obtaining the value of the road surface friction coefficient by optimization calculation so that the sum of squares of the deviation is minimized.

  According to the road surface friction coefficient estimating apparatus according to the present invention, it is possible to always accurately estimate the road surface μ with good response even when traveling on an uphill / downhill.

It is a functional block diagram of a road surface friction coefficient estimating device concerning one embodiment of the present invention. It is a flowchart of the road surface friction coefficient estimation program which concerns on one Embodiment of this invention. 3 is a flowchart continuing from FIG. 2 according to an embodiment of the present invention. It is explanatory drawing of the brush model of the tire which concerns on one Embodiment of this invention. It is explanatory drawing of distribution of the force which acts in the contact surface of the tire which concerns on one Embodiment of this invention. It is explanatory drawing which shows the characteristic of the longitudinal force with respect to the slip ratio which concerns on one Embodiment of this invention. It is explanatory drawing of the weighting coefficient set according to the estimated longitudinal force which concerns on one Embodiment of this invention.

Hereinafter, embodiments of the present invention will be described with reference to the drawings.
In FIG. 1, reference numeral 10 denotes a road surface μ estimation device that estimates a road surface μ, and is connected to a longitudinal acceleration sensor 11, an engine speed sensor 12, an engine control unit 13, and a transmission control unit 14. The engine speed Ne, the engine torque Teg, the main transmission gear ratio i, and the turbine speed Nt of the torque converter are input.

  The road surface μ estimation device 10 calculates an actual longitudinal force generated by the tire at each sampling time as an actual longitudinal force Fmsve according to a road surface friction coefficient estimation program described later based on each input signal described above. The difference between the current value of the force Fmsve and the previous value (previous in the present embodiment) is calculated as an actual longitudinal force difference value ΔFmsve, and the tire model includes the road surface μ as a parameter at the same timing as the actual longitudinal force Fmsve. An ideal longitudinal force generated by the tire is calculated as an ideal longitudinal force Fmdve, and a difference value between the current value of the ideal longitudinal force Fmdve and a past value (previous in the present embodiment) is an ideal longitudinal force difference value ΔFmdve. And the value of the road surface μ is obtained by optimization calculation so that the sum of squares of the deviation between the actual longitudinal force difference value ΔFmsve and the ideal longitudinal force difference value ΔFmdve at least for each sampling time is minimized. Is constructed sea urchin.

  That is, as shown in FIG. 1, the road surface μ estimation apparatus 10 is mainly composed of an actual front / rear force difference value calculation unit 10a, an estimated front / rear force calculation unit 10b, a weight function setting unit 10c, and a road surface μ calculation unit 10d. .

The actual longitudinal force difference value calculation unit 10 a receives the longitudinal acceleration Ax from the longitudinal acceleration sensor 11. Then, for example, the current actual longitudinal force Fmse is calculated by the following equation (1).
Fmse = m · Ax (1)
Here, m is the vehicle mass.

Further, the current actual longitudinal force difference value ΔFmse is calculated by the following equation (2).
ΔFmse = Fmse−Fmse [k−1] (2)

であり、新たに、実前後力差分値ΔＦmseが算出されると、

と更新される。上述のように算出される実前後力差分値ΔＦmsveは路面μ演算部１０ｄに出力される。このように、実前後力差分値演算部１０ａは、実前後力差分値算出手段として設けられている。 Further, when the actual front / rear force difference value calculation unit 10a calculates the current actual front / rear force difference value ΔFmse, the actual front / rear force difference value, which is a vector amount, is calculated from a plurality of actual front / rear force difference values having different sampling times. ΔFmsve is newly set. In the present embodiment, as an example of the actual front / rear force difference value ΔFmsve, ΔFms [0], ΔFms [1],..., ΔFms [m],. A description will be given of a configuration composed of a total of 20 components of ΔFms [19]. That is,

When the actual longitudinal force difference value ΔFmse is newly calculated,

And updated. The actual longitudinal force difference value ΔFmsve calculated as described above is output to the road surface μ calculation unit 10d. Thus, the actual longitudinal force difference value calculation unit 10a is provided as an actual longitudinal force difference value calculation unit.

The estimated longitudinal force calculation unit 10b receives the engine speed Ne from the engine speed sensor 12, receives the engine torque Teg from the engine control unit 13, receives the main transmission gear ratio i from the transmission control unit 14, and the turbine rotation of the torque converter. A number Nt is input. Then, for example, the longitudinal force (estimated longitudinal force) Fme transmitted from the current engine to the tire is calculated by the following equation (5).
Fme = Teg · t · i · η · if / Rt (5)
Here, t is a torque ratio of the torque converter, and is obtained by referring to a preset map of the rotational speed ratio e (= Nt / Ne) of the torque converter and the torque ratio of the torque converter. Also, η is drive system transmission efficiency, if is the final gear ratio, and Rt is the tire radius.

であり、新たに、推定前後力Ｆmeが算出されると、

と更新される。上述のように演算される現在の推定前後力Ｆmeは、重み関数設定部１０ｃ、路面μ演算部１０ｄに出力され、推定前後力Ｆmveは路面μ演算部１０ｄに出力される。 Further, when calculating the current estimated longitudinal force Fme, the estimated longitudinal force calculator 10b newly sets the estimated longitudinal force Fmve, which is a vector amount, based on a plurality of estimated longitudinal forces with different sampling times calculated in the past. In the present embodiment, as an example of the estimated longitudinal force Fmve, Fm [0], Fm [1], ..., Fm [m], ..., Fm [18], Fm [ 19], which is composed of a total of 20 components. That is,

When the estimated longitudinal force Fme is newly calculated,

And updated. The current estimated longitudinal force Fme calculated as described above is output to the weighting function setting unit 10c and the road surface μ calculating unit 10d, and the estimated longitudinal force Fmve is output to the road surface μ calculating unit 10d.

  The weight function setting unit 10c receives the current estimated longitudinal force Fme for each sampling time from the estimated longitudinal force calculation unit 10b. As shown in the following equation (8), the first weight function W1ve, which is a square matrix having the same number of rows and columns as the number of data of the actual longitudinal force difference value ΔFmsve and the estimated longitudinal force Fmve described above, Set and output to the road surface μ calculator 10d.

ここで、第１の重み関数Ｗ1veを構成する各成分は、例えば、図７のマップを参照して設定される値であり、Ｓ／Ｎ比が悪く誤差が大きいと考えられる推定前後力の小さい領域、及び、明らかに誤った値であると考えられる推定前後力の大きな領域では０に設定されるようにして、そのサンプリング時間におけるデータが路面μの推定に影響を与えないようにする。尚、その他にも車速が小さい領域のデータを無効にする、車体すべり角が大きい場合のデータを無効にする（横力が大きいと精度が低下することを考慮）、又は、サンプリング時間の古いデータを無効にする等をこの第１の重み関数Ｗ1veの成分に含ませるようにしても良い。

Here, each component constituting the first weighting function W1ve is, for example, a value set with reference to the map of FIG. 7, and the estimated front-rear force that is considered to have a poor S / N ratio and a large error is small. It is set to 0 in a region and a region having a large estimated longitudinal force that is considered to be an erroneous value so that data at the sampling time does not affect the estimation of the road surface μ. In addition, invalidate the data in the area where the vehicle speed is low, invalidate the data when the vehicle slip angle is large (considering that the accuracy decreases when the lateral force is large), or the data with the old sampling time May be included in the component of the first weighting function W1ve.

  The road surface μ calculator 10d receives the actual longitudinal force difference value Fmsve from the actual longitudinal force difference calculator 10a, the estimated longitudinal force Fmve from the estimated longitudinal force calculator 10b, and the weight function setting unit 10c from the first. The weight function W1ve is input.

First, the main logic of the road surface μ estimation in this embodiment will be described with reference to the tire brush model of FIG. In FIG. 4, with the x axis as the longitudinal direction of the vehicle body, the distance traveled by the grounding point entered from point O (grounding start point) during time Δt, and the x coordinate of point P are
x = u · Δt (9)
It is. Here, u is a vehicle body speed.

In addition, the x coordinate of the point P ′ entered from the point O ′ (the point on the tread base of the contact start point O) is
x ′ = R0 · ω · Δt (10)
It is. Here, R0 is the effective rolling radius of the tire, and ω is the tire rotation speed.

  Accordingly, the relative displacement in the x direction between the point P and the point P ′, that is, the tread rubber deformation is expressed by the following equation (11).

xx ′ = ((u−R0 · ω) / (R0 · ω)) · (R0 · ω) · Δt
= S · x '(11)
Where s is the slip ratio of the tire in the longitudinal direction during driving,
s = (u−R0 · ω) / (R0 · ω) (12)
It is.

Therefore, the force σx in the x direction per unit width and unit length acting on the point P is
σx = −Kx · s · x ′ (13)
It becomes. However, Kx is the longitudinal rigidity of the tread rubber per unit width and length. Here, considering the tire contact pressure distribution,
p = (6 · Fz / b · L) · (x ′ / L) · (1− (x ′ / L)) (14)
Here, b is the contact surface width, and L is the contact surface length.

  Due to the magnitude relationship between the maximum frictional force distributions μp and σ of each part in the contact surface due to contact pressure, the force acting on the contact surface of the tire in the range of the adhesion range represented by 0 ≦ x ′ <x ′s is 13), and expressed in μp in the range of the slip region where x ′ ≧ x ′s. Therefore, the x-direction force acting on the ground contact surface in the adhesive region is σx, and that in the slip region is μp (see FIG. 5).

By the way, substituting the above equations (13) and (14) into σ = μp to obtain x ′s, and a dimensionless display of this is denoted as ξs.
ξs = x ′s / L = 1− (Ks / (3 · μ · Fz)) · λ (15)
It becomes. here,
^{λ = s, Ks = b ·} L 2/2 · Kx ... (16)
It is.

From the above, the force in the x direction (front-rear force Fx) acting on the entire tire contact surface is ξs> 0, that is, when the contact surface is composed of an adhesive region and a slip region,
Fx = b · ∫ (σx) dx (however, the integration range is 0 to xs) (17)
It is.

Also, when ξs ≦ 0, that is, when the ground contact surface is all in the slip region,
Fx = ∫ (−μp) dx ′ (however, the integration range is 0 to L) (18)
The following equations (19) and (20) are obtained by substituting the above equations (13), (14), and (15) into the above equations (17) and (18).

Ξs = 1− (Ks / (3 · μ · Fz)) · λ> 0 Fx = −Ks · s · ξs ² −6 · μ · Fz · ((1/6)
− (1/2) · ξs ² + (1/3) · ξs ³ ) (19)
Here, Fz is a contact load of the tire (for four wheels) (can be approximated by a vehicle body mass m).

Ξs = 1− (Ks / (3 · μ · Fz)) · when λ <0 Fx = −μ · Fz (20)

In contrast to these formulas (19) and (20), formulas (15) to (15) are changed to the following formula (21).
ξs = 1− (Ks / (3 · μ · Fz)) · | s | (21)
Here, assuming that the estimated longitudinal force transmitted from the engine to the tire is Fm and the estimated longitudinal force Fm is a proportional region of Ks (broken line region in FIG. 6), Fm = −Ks · s. Equation (21) becomes the following equation (22).
ξs = 1− (| Fm | / (3 · μ · Fz)) (22)

Similarly, when the above equation (19) is modified, the following equation (23) is obtained.
Fx = Fm · ξs ² −6 · μ · Fz · ((1/6)
− (1/2) · ξs ² + (1/3) · ξs ³ ) (23)
With these equations (22) and (23), the longitudinal force Fx can be handled as an equation of Fm, Fz, and μ. That is, the ideal longitudinal force (ideal longitudinal force) Fmd generated by the tire is obtained by the following equation (24) from the above equations (22) and (23).
Fmd = Fm · ξs ² −6 · μ · Fz · ((1/6)
− (1/2) · ξs ² + (1/3) · ξs ³ ) (24)
However, ξs = 1− (| Fm | / (3 · μ · Fz)) (25)

  Then, the actual longitudinal force difference value ΔFms (= Fms−Fms [k−1]), which is the difference between the current value of the actual longitudinal force Fms and the previous value, and the current value of the ideal longitudinal force Fmd, If the road surface μ in the equations (24) and (25) such that the ideal longitudinal force difference value ΔFmd (= Fmd−Fmd [k−1]), which is a difference value from the previous value, is substantially the same, the road surface μ is calculated. It can be estimated.

  This is to determine an optimum solution as the road surface μ by setting an evaluation function that minimizes the deviation between the actual longitudinal force difference value ΔFms and the ideal longitudinal force difference value ΔFmd and performing a convergence calculation. There is an optimization method as a method for setting an evaluation function that minimizes this deviation and performing a convergence calculation. In this embodiment, the steepest descent method is used to find a solution that minimizes the square error. .

Hereinafter, the evaluation function L [n] is shown in Equation (26). The evaluation function L [n] includes a term for reducing the deviation between the actual longitudinal force difference value ΔFms and the ideal longitudinal force difference value ΔFmd (first evaluation function), and a term for reducing the change amount δμ of the road surface μ estimated value ( (Second evaluation function) (that is, the third evaluation function as a whole).
L [n] = [ΔFmsve−ΔFmdve [n]] ^T W1ve [ΔFmsve−ΔFmdve [n]] + W2 · δμ ²
... (26)

  Specifically, according to the following equation (27), Jacobian Jve [n-1] which is a vector whose element is the amount of change in the ideal longitudinal force difference value ΔFmdve when the road surface μ changes slightly is estimated as the road surface μ. Calculation is performed using the previous value μ [n-1]. The subscript [n-1] of the Jacobian Jve [n-1] represents the previous value μ [n-1] of the estimated value of the road surface μ. When the iterative calculation n-1 = 0, the road surface μ Since there is no previous estimated value μ [n−1], the estimated result μ [z−1] of the road surface μ at the time of the previous sampling is substituted.

ヤコビアンＪve[n-1]の各要素は、以下の各式で求められるものである。

Each element of the Jacobian Jve [n-1] is obtained by the following equations.

(∂ΔFmd [0] / ∂μ [n-1]) = (Fm [0] ² / (Fz · μ [n-1]))
(-1+ (1/27). (Fm [0] / (Fz.μ [n-1])))
− ((Fm [0] [k-1] ² / (Fz · μ [n-1] [k-1]))
(-1+ (1/27) / (Fm [0] [k-1] / (Fz.μ [n-1] [k-1]))))
(∂ΔFmd [1] / ∂μ [n-1]) = (Fm [1] ² / (Fz · μ [n-1]))
(-1+ (1/27). (Fm [1] / (Fz.μ [n-1])))
− ((Fm [1] [k-1] ² / (Fz · μ [n-1] [k-1]))
(-1+ (1/27) / (Fm [1] [k-1] / (Fz.μ [n-1] [k-1]))))
:
(∂ΔFmd [m] / ∂μ [n-1]) = (Fm [m] ² / (Fz · μ [n-1]))
(-1+ (1/27). (Fm [m] / (Fz.μ [n-1])))
− ((Fm [m] [k-1] ² / (Fz · μ [n-1] [k-1]))
(-1+ (1/27) / (Fm [m] [k-1] / (Fz.μ [n-1] [k-1]))))
:
(∂ΔFmd [18] / ∂μ [n-1]) = (Fm [18] ² / (Fz · μ [n-1]))
(-1+ (1/27) / (Fm [18] / (Fz. [Mu] [n-1])))
− ((Fm [18] [k-1] ² / (Fz · μ [n-1] [k-1]))
(-1+ (1/27) / (Fm [18] [k-1] / (Fz.μ [n-1] [k-1]))))
(∂ΔFmd [19] / ∂μ [n-1]) = (Fm [19] ² / (Fz · μ [n-1]))
(-1+ (1/27). (Fm [19] / (Fz.μ [n-1])))
− ((Fm [19] [k-1] ² / (Fz · μ [n-1] [k-1]))
(-1+ (1/27) / (Fm [19] [k-1] / (Fz.μ [n-1] [k-1]))))

  FΔmd [0] to ΔFmd [19] are components of the ideal longitudinal force difference value ΔFmdve (details will be described later), and the subscript [k-1] is the past value of each value (this embodiment) In the form of, the previous value) is shown.

Next, a change amount δμ of the road surface μ estimated value is calculated by the following equation (28).
^{δμ = [Jve [n-1} ] T W1veJve [n-1] + W2] -1
^{Jve [n-1] T W1ve} [ΔFmsve-ΔFmdve [n-1]] ... (28)
Here, W2 is a fixed value determined experimentally.

Next, the road surface μ estimated value μ [n] is calculated by the following equation (29).
μ [n] = μ [n−1] + δμ (29)

  Next, the ideal longitudinal force difference value ΔFmdve [n] is calculated using the road surface μ estimated value μ [n] calculated by the above equation (29). When the number of iterations n = 0, the estimation result at the previous sampling time is substituted.

ここで、ΔＦmdve[n]の各要素は、次式で演算される。 すなわち、前述のタイヤモデルからの（２４）、（２５）式により、
ΔＦmd[0][n]＝Ｆmd[0][n]−Ｆmd[0][n][k-1]
＝（ Ｆm[0]−（Ｆm[0]^２／（Ｆｚ・μ[n]））
＋（１／２７）・（Ｆm[0]^３／（Ｆｚ^２・μ[n]^２）））
−（ Ｆm[0][k-1]−（Ｆm[0][k-1]^２／（Ｆｚ・μ[n][k-1]））
＋（１／２７）・（Ｆm[0][k-1]^３／（Ｆｚ^２・μ[n][k-1]^２））） …（３１）
同様に、ΔＦmd[1][n]〜ΔＦmd[19][n]も演算される。

Here, each element of ΔFmdve [n] is calculated by the following equation. That is, according to the equations (24) and (25) from the above tire model,
ΔFmd [0] [n] = Fmd [0] [n] −Fmd [0] [n] [k−1]
= (Fm [0] − (Fm [0] ² / (Fz · μ [n]))
+ (1/27) · (Fm [0] ³ / (Fz ² · μ [n] ² ))
− (Fm [0] [k-1] − (Fm [0] [k-1] ² / (Fz · μ [n] [k-1]))
+ (1/27) · (Fm [0] [k-1] ³ / (Fz ² · μ [n] [k-1] ² ))) (31)
Similarly, ΔFmd [1] [n] to ΔFmd [19] [n] are also calculated.

  Next, the evaluation function L [n] represented by the above equation (26) is calculated, the previous value L [n-1] of this evaluation function is compared with the current value L [n], and a preset value is obtained. It is determined whether or not it has converged below ε, and if it has converged, the convergence calculation is stopped and the calculated road surface μ estimated value μ [n] is output as the current road surface μ estimated value μ [z]. . If it has not converged below ε, the operation from Jacobian Jve [n−1] is repeated again. Thus, the road surface μ calculation unit 10d functions as an ideal longitudinal force difference value calculation unit and a road surface friction coefficient estimation unit.

  Next, a road surface friction coefficient estimation program executed by the road surface μ estimation apparatus 10 will be described with reference to the flowcharts of FIGS.

  First, in step (hereinafter abbreviated as “S”) 101, necessary parameters, that is, longitudinal acceleration Ax, engine speed Ne, engine torque Teg, main transmission gear ratio i, and turbine speed Nt of the torque converter are read.

  Next, in S102, the actual front / rear force difference value calculation unit 10a calculates the current actual front / rear force difference value ΔFmse according to the above-described equation (2).

  Next, it progresses to S103 and the actual longitudinal force difference value calculating part 10a updates actual longitudinal force difference value (DELTA) Fmsve by the above-mentioned (4) Formula.

  Next, in S104, the estimated longitudinal force calculator 10b calculates the longitudinal force (estimated longitudinal force) Fme transmitted from the current engine to the tire by the above-described equation (5).

  Next, proceeding to S105, the estimated longitudinal force calculation unit 10b updates the estimated longitudinal force Fmve as in the above-described equation (7).

  Next, in S106, the weight function setting unit 10c refers to the map shown in FIG. 7 and the like, and the number of data of the actual front / rear force difference value ΔFmsve and the estimated front / rear force Fmve as shown in the above equation (8). A first weight function W1ve, which is a square matrix having the same number of rows and columns, is set.

  Next, when proceeding to S107, the road surface μ calculating unit 10d is a Jacobian Jve that is a vector whose element is the amount of change in the ideal longitudinal force difference value ΔFmdve when the road surface μ is slightly changed according to the above-described equation (27). [n-1] is calculated using the previous value μ [n-1] of the road surface μ estimated value.

  Next, in S108, the road surface μ calculator 10d calculates the change amount δμ of the estimated value of the road surface μ according to the above-described equation (28).

  Next, proceeding to S109, the road surface μ calculator 10d calculates the road surface μ estimated value μ [n] by the above-described equation (29).

  Next, in S110, the road surface μ calculation unit 10d calculates an ideal longitudinal force difference value ΔFmdve [n] shown in the equation (30) by the above-described equation (31) based on the tire model.

  Next, proceeding to S111, the road surface μ calculator 10d calculates the evaluation function L [n] according to the above-described equation (26).

  In S112, the previous value L [n-1] of the evaluation function is compared with the current value L [n], and whether or not the evaluation function has converged to less than a preset value ε (L [n] −L If [n-1] <ε) is determined, and if it converges, the process proceeds to S113, and the road surface μ estimated value μ [n] is set as the current road surface μ estimated value μ [z] ( μ [z] = μ [n]), if not converged, the process proceeds to S116, and ΔFmdve [n−1] = ΔFmdve [n], μ [n−1] = μ [n], L [n -1] = L [n] is set, and the calculation from S107 is repeated again.

  After setting the current road surface μ estimated value μ [z] in S113, the process proceeds to S114, where the current road surface μ estimated value μ [z] is output, and the process proceeds to S115, where the current road surface μ estimated value μ [z] is output. [z] is updated to the previous road surface μ estimated value μ [z-1] (μ [z-1] = μ [z]), and the program exits.

  As described above, according to the embodiment of the present invention, the actual longitudinal force generated by the tire at each sampling time is calculated as the actual longitudinal force Fmsve, and the current value and the past value of the actual longitudinal force Fmsve are calculated. Is calculated as the actual front / rear force difference value ΔFmsve, and the ideal front / rear force Fmdve is calculated as the ideal front / rear force Fmdve by the tire model including the road surface μ as a parameter at the same timing as the actual front / rear force Fmsve. The difference value between the current value of the longitudinal force Fmdve and the past value is calculated as an ideal longitudinal force difference value ΔFmdve, and at least the square of the deviation between the actual longitudinal force difference value ΔFmsve and the ideal longitudinal force difference value ΔFmdve for each sampling time. The value of the road surface μ is obtained by optimization calculation so that the sum is minimized. For this reason, even when traveling on an uphill / downhill, errors caused by this are removed by differential processing, and it is possible to always estimate the road surface μ with high accuracy with good response.

  In this embodiment, the convergence determination of the evaluation function L [n] is performed until it becomes less than ε. However, the number of convergence calculations may be set in advance. Also, a limit value for the number of operations may be provided.

  Furthermore, in this embodiment, when calculating the difference value with the past value, the difference between the current value and the previous value is calculated. The difference may be calculated.

DESCRIPTION OF SYMBOLS 10 Road surface micro-estimation apparatus 10a Real front-back force difference value calculating part (Actual front-back force difference value calculation means)
10b Estimated longitudinal force calculation unit 10c Weight function setting unit 10d Road surface μ calculation unit (ideal longitudinal force difference value calculation means, road friction coefficient estimation means)
11 Longitudinal Acceleration Sensor 12 Engine Speed Sensor 13 Engine Control Unit 14 Transmission Control Unit

Claims (3)

Calculate the actual longitudinal force generated by the tire at each sampling time as the actual longitudinal force, and calculate the difference between the current actual longitudinal force and the previous value as the actual longitudinal force difference value. Difference value calculating means;
The ideal longitudinal force generated by the tire is calculated as an ideal longitudinal force by a tire model including the road friction coefficient as a parameter at the same timing as the actual longitudinal force, and the current value and the past value of the ideal longitudinal force are calculated. An ideal longitudinal force difference value calculating means for calculating the difference value as an ideal longitudinal force difference value;
Road surface friction coefficient estimating means for obtaining the value of the road surface friction coefficient by optimization calculation so that the sum of squares of the deviation between the actual front / rear force difference value and the ideal front / rear force difference value at least for each sampling time is minimized;
A road surface friction coefficient estimating device comprising:   The road surface friction coefficient estimating means is a first evaluation function obtained by multiplying a value obtained by squaring the deviation at each sampling time by a weight function according to a measurement condition at each sampling time, and a previously calculated road surface friction coefficient. A second evaluation function including a value obtained by squaring the correction amount of the road surface friction coefficient this time, and a third evaluation function that is the sum of the first evaluation function and the second evaluation function are obtained, 2. The current road surface friction coefficient is obtained by calculating a correction amount of the road surface friction coefficient using the fact that a value obtained by partial differentiation of the evaluation function of 3 with respect to a road surface friction coefficient is 0. Road friction coefficient estimation device.   3. The road surface friction coefficient estimating apparatus according to claim 2, wherein the weight function is set according to a longitudinal force transmitted from the engine to the tire.

Images