JP3329171B2 - Rotation speed detector - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an apparatus for detecting a rotation speed, and more particularly to a technique for removing a periodic disturbance that has entered the rotation speed.

[0002]

2. Description of the Related Art Wheel speed is used in a vehicle control device for controlling the motion of a vehicle by controlling the motion state of the wheel, for example, the slip ratio of the wheel and the braking / driving force. further,
It is also used in a wheel characteristic detecting device that detects a characteristic of a wheel such as a tire pressure of the wheel.

A device for detecting a wheel speed generally includes a pulse signal generating circuit for detecting the rotation of a rotating body that rotates together with the wheel and generating a pulse signal corresponding to the wheel speed, and a periodic signal generated based on the generated pulse signal. And a wheel speed determination circuit for determining a wheel speed.

A pulse counting device used in a wheel speed determining circuit of such a wheel speed detecting device is disclosed, for example, in Japanese Patent Application Laid-Open No. 60-93824. This device divides the gate time for counting pulses into multiple parts, counts the number of pulses counted at each time of the time of the multiplied gate time, stores it in the first register, and shifts to the register at the subsequent stage every time. The configuration is

[0005]

Since the conventional apparatus has a structure in which the counted number of pulses is shifted by a plurality of registers every time the time elapses, a number of registers corresponding to the division number of the gate time is required. It is difficult to reduce the memory capacity for setting, and there is a problem that a large memory capacity is required.

[0006] The present invention has been made in view of the above points,
Provided is a rotation speed detection device that calculates a derived number from the number of periodic pulse signal edges and the previous carrier and calculates a carrier from the number of edges and the derived number to reduce the number of registers and reduce the memory capacity. The purpose is to do.

[0007]

According to the first aspect of the present invention, a rotation of a rotator is detected to generate a periodic pulse signal corresponding to the rotation speed of the rotator, and a pulse signal is generated based on the period of the pulse signal. The rotational speed is calculated, and among the plurality of correction values previously determined and stored in the plurality of registers, a correction value having a predetermined relationship with the currently calculated rotational speed is read, and the rotational speed calculated this time is calculated. A rotation speed detecting device for determining a rotation speed from which the periodic disturbance with respect to the rotating body has been removed, wherein an edge counting means for counting the number of edges of the pulse signal in a predetermined time period; and deriving the number of the corresponding integer part of the value obtained by dividing the added value by a predetermined number of the carrier to store the number of the rotational speed in the current of a predetermined time period in the cycle
And deriving the number of calculation means for calculating Te, or the sum of the carrier in a predetermined time period that this counted number of edges and the previous
And a carrier calculating means for calculating a carrier by subtracting the derived number and the predetermined number of multiplied values, and reading out rotational speeds corresponding to the derived number from the plurality of registers to eliminate a periodic disturbance. It is used to determine the rotation speed, and updates the rotation speed of the register from which the above reading was performed.

For this reason, it is sufficient to prepare as many registers as the number obtained by dividing the number of edges obtained in one rotation of the rotating body by a predetermined number, and the number of registers can be greatly reduced from the number of registers required conventionally. Therefore, the memory capacity can be significantly reduced.

[0009]

FIG. 2 is a block diagram showing an embodiment of the apparatus according to the present invention. In the figure, 10 is a rotor, and 12 is an electromagnetic pickup. The rotor 10 rotates together with the wheels 14 shown in FIG. 3, and has a number of teeth 16 on the outer circumference. The electromagnetic pickup 12 generates a voltage that changes periodically as the teeth 16 pass. This voltage is shaped into a rectangular wave by the waveform shaper 18 and supplied to the I / O port 22 of the computer 20. There are four wheels 14, and each electromagnetic pickup 12
Are all connected to a computer 20 via a waveform shaper 18, but only one set is typically shown in FIG.
That is, in the present embodiment, each electromagnetic pickup 1
2 and the waveform shaper 18 cooperate with each other to constitute an example of a periodic signal generator of a type that outputs a rectangular wave voltage signal as a periodic signal.

The wheel 14 is, as shown in FIG.
4 is a tire-equipped wheel having a tire 26 mounted on the outer periphery thereof. As shown in FIG. 4, it is considered that a rim side portion 28 and a belt side portion 30 that are relatively rotatable are connected by a torsion spring 32. Can be. Since the rotor 10 is mounted so as to rotate integrally with the wheel 24, the electromagnetic pickup 12 detects the angular velocity of the rim side portion 28 strictly.

As shown in FIG. 2, the computer 20 has a CPU 40 as a processing device and a ROM as a first storage device.
42 and a RAM 44 as a second storage device. The ROM 42 stores a control program represented by the flowchart of FIG. 5 to constitute a rim side rotation speed calculation / correction unit 45 shown in FIG. I have. This computer 20 is connected to another computer 47. This computer 47 has a CPU as a processing device.
48, a ROM 49 as a first storage device, a RAM 50 as a second storage device, and an I / O port 51 as an input / output device. The ROM 49 includes a routine shown in the flowcharts of FIGS. By storing various control programs, the disturbance observer 52, the pre-processing unit 54, the correlation operation unit 56,
The normalization unit 58, the constant correction unit 60, the determination unit 62, and the wheel speed output unit 64 are configured.

As shown in FIG. 2, a display device 66 for notifying the driver of the determination result of the determination section 62 is connected to the I / O port 51 of the computer 47. The display device 66 is a liquid crystal display in the present embodiment, but it is also possible to use another display device such as a lighted or flashing lamp, and various forms including a voice notification device that notifies the driver by voice. It is possible to employ a notification device.

The I / O port 51 of the computer 47 is further provided with a driving / braking torque for detecting a driving / braking torque applied to the wheel 24 (rim side portion 28) by a strain gauge or the like attached to the shaft of the wheel 24. The detection device 68 is connected. The rim side rotation speed calculation / correction unit 45 determines each wheel 14 based on a rectangular wave signal (hereinafter also referred to as a pulse signal) supplied from each of the electromagnetic pickups 12 and the waveform shaper 18 corresponding to the four wheels 14. Is calculated, and the rotation speed of each wheel 14 is corrected by a signal delay / superposition process described later. Manufacturing and assembly errors are present in each wheel 14 and rotor 10, and periodic errors in the rotational speed are generated due to these errors and the like. Ask for. FIG. 6 shows the frequency characteristics (frequency and power) of the rotational speed obtained under the running condition where the vehicle speed is 120 [km / h], the tire radius is about 330 [mm], and the tire pressure is 0.2 [MPa]. (Relationship with spectrum) is shown by a graph. As is clear from the graph, a plurality of harmonics having a fundamental frequency of about 16 [Hz] are generated, which is a natural rotational speed error of the wheel 14, that is, a periodic disturbance. The correction is made. That is, in the present embodiment, the inherent rotational speed error of the wheel 14 is a periodic disturbance to be removed.

The rotational speed of the wheel 14 is calculated based on the peripheral speed. For this purpose, a substantial radius of the tire 26 (the distance from the road surface to the center of the wheel 14 when the tire is deformed by the load) is required. Which depends on the air pressure of the tire 26. Therefore, at first, the normal radius when the air pressure is normal is used. However, if the air pressure change of the tire 26 is found by the processing described later, the air pressure change is obtained from the tire diameter table stored in the ROM 42 in advance. Is read and used. When the rotation speed is calculated based on the angular speed, the tire diameter table is not required.

The function of the rim side rotation speed calculation / correction unit 45 is fulfilled by executing the rotation speed calculation / correction routine shown in FIG. 5 and the disturbance elimination filter routine shown in FIG. 13. First, the principle will be described. The disturbance entering the rotation speed has a periodicity of making one rotation with the wheel 14. Therefore, by utilizing this periodicity, the periodic disturbance is removed by delaying the preceding part of the temporal change in the rotational speed by exactly one cycle and superimposing it on the subsequent part.

The transfer function of the signal delay / superposition process can be represented by a block diagram in FIG. 7, and the state equation in this case is obtained by using the z variable.

[0017]

(Equation 1)

## EQU1 ## Here, v: a rotation speed obtained by cutting a DC component of the input rotation speed v ₀ y: a rotation speed as an output L: a delay amount described in units of a sampling period T of the rotation speed v, where z ⁻¹ = e ^{− j} ω _n . Here, j: imaginary unit ω _n : normalized angular frequency Note that the normalized angular frequency ω _n is not the actual angular frequency ω _A described in units of 1 second, but the angular frequency described in units of the sampling period T. It is. Therefore, the normalized angular frequency omega _n is represented by ω _n = ω _A T.

As a result, the frequency response H (ω) becomes

[0020]

(Equation 2)

Is obtained. As is apparent from this equation, when ω _n L = (2n + 1) π (n is an integer), | H
(Ω) | = 1. As a result, ω _n L = (2n + 1)
When π, y = v.

For example, when the vehicle travels at a vehicle speed of 120 km / h, a disturbance having a harmonic component having a fundamental frequency of 16 Hz exists in the wheel speed as shown in FIG. By the way, in the equation (1), when ω _n L = 2πn, the gain characteristic | H (ω) | has a minimum peak. Since the purpose of this filter is to remove harmonics having a fundamental frequency of 16 Hz, the delay time L may be selected so that the fundamental frequency matches the peak of the filter.

Here, if the fundamental frequency is f ₁ , f ₁
Is

[0024]

(Equation 3)

The following relationship is satisfied. ω _n L = 2πn and n = 1
In this case, the peak of the filter is made to coincide with the fundamental frequency f ₁ , that is, it may be equivalent.

[0026]

(Equation 4)

If T = 5 msec and f ₁ = 16 Hz, L = 12, and the frequency response at that time is shown in FIG.
6 and 8, it can be seen that the peak of the disturbance coincides with the peak of the filter, and the disturbance can be removed. In the graph of FIG. 8, the frequency response H (ω) is expressed in dB on the vertical axis, so that the frequency response H (ω) is 1 is represented by 0 in the graph, and the frequency response H (ω) ) Is 0 in the graph. Therefore, as is apparent from the graph, the frequency response H (ω) becomes sufficiently close to 0 at each position of the frequency at which a plurality of harmonics having a fundamental frequency f _{1 of} 16 [Hz] is generated, and the other frequencies At the position, the frequency response H (ω) is sufficiently close to one.

Next, a description will be given of a specific technique for implementing a rotational speed calculation / correction technique using a computer according to this principle. First, a brief description will be given. Each time each pulse (each unit wave constituting a series of rectangular waves) is supplied from the electromagnetic pickup 12, the original rotation speed v is sequentially calculated based on the time interval between each pulse. The time interval of each pulse is, for example, the edge interval time of each pulse.

Every time one sampling period T elapses, the average value of at least one of the original rotation speeds v ₀ input and calculated during that period is calculated as the sample rotation speed v. Further, a sample fluctuation rotation speed Δv (corresponding to “v” in FIG. 7) is obtained by a high-pass filter having a cutoff frequency of 1 Hz.

The RAM 44 is provided with a memory for one rotation. The memory for one rotation stores an error value E (corresponding to E in FIG. 7) in association with, for example, each tooth of the rotor 10. The error value E is a periodic disturbance to be removed,
That is, in the present embodiment, the value is a value indicating a state in which the disturbance caused by the non-uniformity of the wheels changes with the rotation of the rotor 10, and the acquisition method will be described later in detail. The memory for one rotation stores the number of pulses output from the electromagnetic pickup 12 during one rotation of the rotor 10, that is, the rotor 10.
The same number of storage addresses as the number of teeth (for example, 48) are provided, and each error value E is sequentially stored in each storage address.

FIG. 9 conceptually shows the structure of the memory for one rotation, taking as an example the case where the rotor 10 has eight teeth. In the figure, “1-1” indicates an error value E calculated first during the first rotation, and “1-2” indicates an error value E calculated second during the first rotation. , "2-1" indicates the error value E calculated first during the second rotation, and the left-hand number of the two numbers connected by a hyphen indicates the number of rotations of the wheel 14 , And the number on the right side indicates the number of each pulse in each rotation.

Every time the sampling period T elapses, the error value E is read from the memory for one rotation. At least one error value E already stored in the memory for one rotation
Among them, the error value group E in which each of the at least one variable rotational speed Δv relating to the current sample variable rotational speed Δv and the pulse generation rotational position of the rotor 10 are common is read.
That is, the error value group E obtained just before the rotation of the rotor by one rotation from the current sample fluctuation rotation speed Δv is read as the old error value group E. An average value is obtained for the read old error value group E, and the average old error value E _MEAN is obtained.
It is said.

For example, referring to the example of FIG. 9, assuming that two variable rotation speeds Δv are obtained in each sampling, as shown in FIG.
“1-1” and “1-2” in FIG. 9 are “1-”,
That is, the first old error value group E during the first rotation
And the average value of the two old error values E respectively indicated by “1-1” and “1-2” is set as the average old error value E _MEAN indicated by “1-”. Further, assuming that four variable rotation speeds Δv are obtained in each sampling, for example, as shown in FIG.
"1" to "1-4" constitute "1-", that is, a first old error value group E at the time of the first rotation, and "1-1".
The average value of the four old error values E respectively indicated by .about. "1-4" is defined as the average old error value _E.sub.MEAN indicated by "1-".

The product of the current average old error value E _MEAN thus obtained and the gain g is obtained.
After the sum with the current sample fluctuation rotation speed Δv is calculated, the product of (2-g) / 2 is calculated, and the result is the output y in FIG. That is, the output y is obtained as (old E _MEAN · g + v) · (2-g) / 2.

Next, a new error value E is obtained from the current output y and the current average old error value E _MEAN . That is, the new error value E is obtained as E = old E _MEAN + y · 2 / (2-g). Here, the operation of calculating the output y is an example of the output feedback processing.

When the new error value E is obtained, the error value E is updated in the memory for one rotation. That is,
Each of the old error values E constituting the current old error value group E is rewritten to the same new error value E. FIG.
To explain using an example of 0, for example, when a new error value E corresponding to “2-” is calculated, “1-1” constituting the old error value group E indicated by “1-” And "1-2"
Are rewritten to the same new error value E.

It should be noted that the error value E stored in the memory for one rotation is obtained by repeating the above processing many times.
Does not accurately reflect the periodic disturbance to be removed, the output y becomes 0, and thus the new error value E becomes the old error value E. At this stage, the error value E is rewritten. Also, the error value E is substantially fixed.

Here, the relationship between the memory for one rotation and the delay amount L will be described. In this embodiment, the calculation of the original rotation speed v is performed every time each pulse is generated. However, the calculation of the average error value E _MEAN , the calculation of the output y, and the calculation of the final rotation speed v are not performed for each pulse. No sampling period T
(For example, 6 msec). Therefore, 1
The rotation memory stores substantially the same number of error values E (FIG. 1) as the number of pulses sampled during one rotation of the wheel 14.
It can be said that this is a memory that stores four (in the example of 0, two in the example of FIG. 11). Further, in the present embodiment, the amount of delay when the previously calculated original rotational speed v is delayed by one rotation of the rotor and superimposed on the originally calculated original rotational speed v is stored in the memory for one rotation. Is equal to the number of error value groups E (ie, the number of sample rotation speeds v).

In a unit of the sampling period T, L = 2πR / V / TR, the tire radius V is represented by the vehicle speed, and the sampling period T is unchanged.
Therefore, in order to remove the periodic disturbance using its periodicity, it is necessary to change the delay amount L according to the vehicle speed V, that is, the rotation speed v. On the other hand, in the present embodiment, as described above, the number of memories for one rotation is equal to the number of teeth of the rotor 10, and the error value E for the latest one rotation of the rotor is always stored. You. In addition, every time the sampling period T elapses, an average value of at least one error value E calculated during the sampling period T is obtained and used as the only representative value, so that the storage address of the memory for one rotation is substantially obtained. Is divided according to the number of the original rotational speeds v obtained for each sampling period T.
The number of divisions is equal to the delay amount L. Therefore, in the present embodiment, the delay amount L is automatically changed in accordance with the rotation speed v without performing a delay calculation having a different delay amount L for each rotation speed v.

After the output y from which the periodic disturbance has been removed is calculated as described above, the final rotational speed v from which the periodic disturbance has been removed can be obtained by adding the offset value to the output y. Is calculated. FIG. 12 is a graph showing two examples of the temporal change of the variable rotation speed Δv. The upper graph shows an example of a temporal change when the rotation speed v is slow, and the lower graph shows an example of a temporal change when the rotation speed v is fast. Comparing these graphs, it can be seen that the higher the rotation speed v, the larger the number of variable rotation speeds Δv stored in the memory for one rotation during one sampling (indicated by 図 in the figure). It can also be seen that the higher the speed v, the smaller the number of variable rotational speed groups Δv corresponding to one delay amount L.

While the contents of the rotation speed calculation / correction technique have been described briefly above, a specific description will now be given based on the rotation speed calculation / correction routine of FIG. First, step S
At 51 (hereinafter, simply represented by S51; the same applies to other steps), the value of the integer i is initialized to 1, and then at S52, a state of waiting for the start of the current sampling period is entered. If the current sampling cycle has started, it is determined in S53 whether or not the current sampling cycle has ended. Since this time has just been started, the determination is NO, and the process waits for the rising and falling edge detection pulses of the pulse (wheel speed pulse) output from the electromagnetic pickup 12 to be output in S54. The edge detection signal of the wheel speed pulse is supplied from, for example, an ABS (anti-lock brake system). If it is issued, the determination becomes YES, the duration of the pulse is measured in S55, and the i-th original rotational speed v _(i) is calculated based on the duration in S56, and the calculated value is stored in the RAM 44. Is stored in the memory for one sampling.

Thereafter, in S57, the value of the integer i is 1
It is increased, in S58, the current value of the integer i whether exceeds the maximum value i _MAX is determined. Since the maximum value i _MAX is set to twice the number of teeth of the rotor 10 (corresponding to the rising and falling edge detection pulses of the wheel speed pulse), this determination is ultimately made for the original rotation speed v ₍ Every time _i) is obtained, the result is YES. This time the integer i
Is determined not to exceed the maximum value i _MAX , the determination is NO, and the process immediately returns to S53. However, if it is determined that the current value is exceeded, the determination is YES, and the value of the integer i is reset to 1 in S66. After that, the process returns to S53.

Thereafter, if the current sampling period is completed while the execution of S53 to S58 and S66 is repeated, the determination in S53 is YES, and the process proceeds to the disturbance removal routine in S60. FIG. 13 shows a flowchart of the disturbance removal routine. In the figure, S1 which is a derived number calculating means
At 00, the previous carrier is added to the number of rising and falling edge detection pulses of the wheel speed pulse supplied from the ABS within the immediately preceding predetermined time (6 msec), that is, the number of input edges, and this added value is used as a coefficient (predetermined number). ) Let the integer part of the value divided by N (N is an integer of 2 or more) be the derived pulse number (derived number). Next, in step S102, the number of input edges and the number of pulses derived from the sum of the previous carrier are calculated by a factor N in S102.
Is subtracted to obtain the previous carrier. By the way, M is the number of registers to be used. When the number of teeth of the rotor 10 is 48, M = 48 for N = 2 and M = 48 for N = 4.
24, N = 8, M = 12.

Next, in S104, it is determined whether or not the number of derived pulses exceeds 0. If the number of derived pulses is 0, the process is terminated. If the number of derived pulses is not 0, the process proceeds to S106. In S106, 0 is set to a working buffer (buffer), the current register number n is set to a variable j,
Reset the variable i to 0. Thereafter, in step S108, the variable i
Is smaller than the number of derived pulses, and if i <the number of derived pulses, the register (regi) designated by the variable j in S110.
ster [j]) is added to the value of the buffer and stored in the buffer again. Thereafter, in S112, it is determined whether or not the variable j is smaller than the number M of registers. If j <M, j is determined in S114.
Is incremented by one, and if j ≧ M, the value of j is reset to zero. This is for sequentially incrementing the value of the variable j between 0 and M-1 because the number of registers is M. After the execution of S114 or S116, the process proceeds to S118, where the variable i is incremented by 1, and the process proceeds to S108. On the other hand, if i ≧ the number of derived pulses in S108, the process proceeds to S120.

Here, if N = 2 and M = 48, the number of registers is 48, the same as the number of teeth of the rotor, and the registers correspond to the teeth of the rotor on a one-to-one basis. The stored rotation speed is stored. S106 and above
In S118, the sum of the rotational speeds for the number of derived pulses calculated within the predetermined time of 6 msec is stored in the buffer.

At S120, the total of the rotational speeds stored in the buffer is divided by the number of derived pulses to obtain an average value of the rotational speeds for a predetermined time of 6 msec, and this is used as a register output. This register output corresponds to the average old error value E _MEAN in FIG. In the next S122, a new error value E is obtained by multiplying the register output by the gain g and adding the result to the input wheel speed v passed through the high-pass filter.

Next, the variable i is reset to 0 in S124. Thereafter, in S126, it is determined whether or not the variable i is smaller than the number of derived pulses. If i <derived pulse, the new error value is subtracted from the register output in S128, and this is replaced by the variable n representing the current register number. Register (re
gister [n]).

Thereafter, in step S130, the variable n is set to the number of registers M.
It is determined whether or not n is less than n. If n <M, n is incremented by 1 in S132, and if n ≧ M, S134 is performed.
Resets the value of n to 0. This is because the current register number is sequentially incremented between 0 and M-1 because the number of registers is M. S132 above,
After execution of S134, the process proceeds to S136, in which the variable i is incremented by 1, and the process proceeds to S126. On the other hand, in S126, i
If ≧ the number of derived pulses, the new error value is multiplied by the gain (2-g) / 2 in S138, and a filter output (output y in FIG. 7)
Is obtained, and the process ends.

In steps S124 to S136, the rotational speeds stored in the registers for the number of derived pulses are updated within a predetermined time of 6 msec. FIG. 14 shows the predetermined time 6 msec, N =
2 shows how the wheel speed pulse, the number of edges of the wheel speed pulse, the number of pulses obtained from the number of edges, and the number of derived pulses change when M = 48. In this case, the rotational speeds V _{1 to} V ₉ calculated within the predetermined time are sequentially stored in 48 registers corresponding to each tooth of the rotor 10. For this reason, the number of edges obtained in one rotation of the rotating body is determined by the coefficient N
It is sufficient to prepare the number of registers divided by the (predetermined number), and the number of registers can be greatly reduced from the number of registers required for the edge, which has been conventionally required, and therefore, the memory capacity can be significantly reduced.

FIG. 1 shows the frequency response of the gain and the phase of the disturbance elimination filter when N = 2 and M = 48 obtained from the actual measurement data of the vehicle running at a speed of 90 km / h.
5 (A) and (B). Further, FIGS. 15C and 15D show the frequency response of each of the gain and the phase of the disturbance elimination filter when N = 4 and M = 24. As is clear from this, even if the number of registers is reduced from 48 to 24, there is no significant change in the frequency response of each of the gain and phase of the disturbance elimination filter, and the performance of the disturbance elimination filter hardly deteriorates. That is, the number of registers can be reduced, that is, the memory capacity can be reduced without performance degradation.

Next, the disturbance observer 52 will be described. The disturbance observer 52 is configured based on the model of the wheel 14 shown in FIG. Hereinafter, the configuration of the disturbance observer 52 will be described. The wheels 14, when modeled as being connected by a torsion spring 32 of the belt side 30 Togabane constant K of inertia J _R of the rim 28 and moment of inertia _{J B, (2) ~ (} 4) Is established, which constitutes a linear system.

_JR ω _R ′ = −Kθ _RB + T ₁ (2) J _B ω _B ′ = −Kθ _RB −T _d (3) θ _RB ′ = ω _R −ω _B. (4) where ω _R : angular velocity of the rim side part ω _R ': angular acceleration of the rim side part ω _B : angular velocity of the belt side part 30 ω _B ': angular acceleration of the belt side part 30 _RB : rim side The torsion angle T ₁ between the portion 28 and the belt side portion 30: the driving / braking torque detected by the driving / braking torque detecting device 68 T _d : the disturbance torque from the road surface Note that the rim side portion 28 and the belt side portion are actually used. Although there is a damper between it and 30, its influence is relatively small,
In the present embodiment, its existence is ignored.

If the above state equation is expressed using a vector and a matrix, the equation (5) is obtained.

[0054]

(Equation 5)

Here, the movement of the wheel 14 when the air pressure of the tire 26 changes and the spring constant of the torsion spring 32 changes from K to K + ΔK is expressed by equation (6).

[0056]

(Equation 6)

That is, a change in the spring constant K by ΔK is equivalent to a disturbance represented by the last term on the right side of the equation (6) being applied to the normal tire 26. Since the disturbance includes information on the change amount ΔK of the spring constant K, and the spring constant K changes in accordance with the air pressure of the tire 26, the change in the tire air pressure is estimated by estimating the disturbance. Can be estimated. The disturbance observer technique is used to estimate the disturbance. If the torque _Td from the road surface is also treated as a disturbance, the disturbance w to be estimated is expressed by the following equation (7).

[0058]

(Equation 7)

However, theoretically, only one element of the disturbance [w] can be estimated, and therefore the second element w ₂ is estimated. By defining the disturbance w ₂ in (8), for the state equation of the wheel 14 is made to equation (9), constitutes a disturbance observer based on the equation (9).

W ₂ = (− 1 / J _B ) T _d + (ΔK / J _B ) θ _RB (8)

[0061]

(Equation 8)

The disturbance observer estimates disturbance as one of the state variables of the system. Therefore, (8) for inclusion in the state of the disturbance w ₂ system equation to approximate the disturbance dynamics to be estimated by equation (10). w ₂ ′ = 0 (10) This means that the continuously changing disturbance is approximated stepwise (zero order approximation) as shown in FIG. 16, and the disturbance estimation speed of the disturbance observer 52 is estimated. This approximation is well tolerated if it is fast enough compared to the change in disturbance to be made. (1
From equation (0), if the disturbance w ₂ is included in the system state, (1)
The extension system of the expression 1) is configured.

[0063]

(Equation 9)

In equation (11), [w _B θ _RB
w ₂ ] ^T cannot be detected. Therefore, if the disturbance observer 52 is configured based on this system, the disturbance w ₂ and the state variables ω _B , θ that cannot be measured originally
_RB can be estimated. In order to simplify the description, the vectors and matrices in equation (11) are decomposed and expressed as follows.

[0065]

(Equation 10)

At this time, the state [Z] = [ω _B θ _RB
The configuration of the minimum dimension observer for estimating w ₂ ] ^T is expressed by equation (12). [Z _p '] = [A ₂₁ ] [X _a ] + [A ₂₂ ] [Z _p ] + [B ₂ ] [u] + [G] ｛[X _a ′] − ([A ₁₁ ] [X _a ] + [A ₁₂ ] [Z _p ] + [B ₁ ] [u])｝ = ([A ₂₁ ] − [G] [A ₁₁ ]) [X _a ] + ([A ₂₂ ] − [G] [ _{A 12]) [Z p]} + [G] [X a '] + ([B 2] - [G] [B 1]) [u] ··· (12) However, [Z _p]: [Z ] [Z _p ']: Rate of change of the estimated value [Z _p ] [G]: Gain that determines the estimated speed of the disturbance observer 52 This equation is represented in a block diagram as shown in FIG. In the figure, [I] is a unit matrix, and s is a Laplace operator.

The error [e] between the true value [Z] and the estimated value [Z _p ] is set as [e] = [Z] − [Z _p ], and the rate of change of the error [e] is [e ′]. ], The relationship of Expression (13) is obtained. [E ′] = ([A ₂₂ ] − [G] [A ₁₂ ]) [e] (13) This represents the estimated characteristic of the disturbance observer 52,
Matrix - a _{([A 22] [G]} [A 12]) eigenvalues i.e. pole of the disturbance observer 52. Therefore, the estimated speed of the disturbance observer 52 increases as the eigenvalue moves away from the origin on the left half surface of the s-plane. The observer gain [G] may be determined so as to achieve a desired estimated speed.

In the above description, the disturbance w ₂ is calculated by the equation (8).
That is, w ₂ = (− 1 / J _B ) T _d + (ΔK / J _B ) θ
The disturbance w when the spring constant K of the torsion spring 32 of the observer 52 changes by ΔK is represented by _RB.
Although the configuration of the portion for estimating ₂ has been described, the inertia moment J _B of the belt side portion 30 of the disturbance observer 52 is J _B +
If changes to .DELTA.J _B, as well as the moment of inertia J _R of the rim 28 is constructed in the same manner portion for estimating respective disturbances when changed to J _R + ΔJ _R.

The pre-processing unit 54 controls the correlation
This is a part for performing pre-processing of the operation. Rim side 2 detected
8 angular velocity ω_REstimated by the disturbance observer 52
The estimated angular velocity ω of the belt side 30_BpAnd angular acceleration ω
_R'And the angular acceleration estimate ω _Bp'Is required.

A correlation operation is performed in a correlation operation unit 56 using the disturbance w _2p , angular velocity ω _R , ω _Bp , angular acceleration ω _R ', ω _Bp ', torsion angle θ _{RBp, and the} like. Then, a change in the spring constant K of the torsion spring 32 is determined. First, the correlation calculation unit 56 executes a correlation calculation routine for acquiring a spring constant change, which is represented by the flowchart in FIG.

In the initial setting of S21, the integer i is set to 1
Is reset, and the disturbance w expressed by the equation (8) is reset._TwoGuess
Constant value w_2pAnd torsion angle estimate θ_RBpCross-correlation C with
(W_2p, Θ _RBp) And estimated torsion angle θ_RBpAutocorrelation C
(Θ_RBp, Θ_RBp) Are reset to 0. RAM5
Zero cross-correlation memory and auto-correlation memory contents become zero
It is done.

Subsequently, in S22, the current estimated disturbance value w
_{2p (i)} and the estimated torsion angle θ _{RBp (i)} are read and S
At 23, the product of the disturbance estimation value w _{2p (i)} and the torsion angle estimation value θ _{RBp (i)} is calculated and added to the cross-correlation C (w _2p , θ _RBp ). However, when S23 is first executed, the cross-correlation C (w _2p , θ _RBp ) is 0, and therefore, the estimated disturbance w _{2p (i)} and the estimated torsion angle θ _{RBp (i) are} stored in the cross-correlation memory. Is merely stored.

Similarly, in S24, the square of the estimated torsion angle θ _{RBp (i)} is calculated, and the autocorrelation C in the autocorrelation memory is calculated.
(Θ _RBp , θ _RBp ). In S25, it is determined whether or not the integer i is equal to or greater than the predetermined integer M. Since the determination is initially NO, the integer i is determined in S26.
Is increased by 1, and S22 to S24 are executed again.

When this execution has been repeated M times, the determination in S25 becomes YES, and one execution of the correlation calculation routine for obtaining a spring constant change ends. After the cross-correlation C (w _2p , θ _RBp ) and the auto-correlation C (θ _RBp , θ _RBp ) are obtained in the correlation operation unit 56 as described above, the normalization unit 5
In step 8, the L _K value is obtained from equation (21),
It is stored in 50 L _K value memories.

L _K = C (w _2p , θ _RBp ) / C (θ _RBp , θ _RBp ) (21) The L _K value is expressed by the equation (22) based on the equation (8). L _K = (− 1 / J _B ) C ₀ + ΔK / J _B (22) where C ₀ is C (T _dP , θ _RBp ) / C (θ _RBp , θ
_RBp ), which is independent of the change in the spring constant K, and can be compensated for by _obtaining the tire pressure in a normal state in advance. Also, C
(T _dP , θ _RBp ) represents a cross-correlation between the estimated value of the disturbance torque T _{d and} the estimated value of the torsion angle θ _RB .

In the constant correction section 60, L _K = C obtained as described above and stored in each L-value memory.
The spring constant K of the torsion spring 32 is corrected based on (w _2p , θ _RBp ) / C (θ _RBp , θ _RBp ). Since L _K is represented by L _K = (− 1 / J _B ) C ₀ + ΔK / J _B as described above, the relationship between L _k and ΔK is stored in advance in the ROM 49 as a spring constant change table. The spring constant change amount ΔK is obtained based on the table, and the spring constant K of the disturbance observer 52 is corrected by the change amount.

When the disturbance observer 52 is operated for the first time after the key switch of the vehicle is turned on, regular values are used as the spring constant K, the inertia moment J _R and the inertia moment J _B. If so, the corrected value is used as the spring constant K. Therefore, the spring constant change amount ΔK obtained in that state is a correction amount from the corrected value.

However, since the determination unit 62 requires a change amount from the normal value, when the key switch is turned on, the spring constant correction value memory is cleared, and the correction is performed by the constant correction unit 60. Each time, the correction value ΔK is added to the contents of the memory. In the determination section 62, the correction value ΔK stored in the spring constant correction value memory is stored in the ROM.
This is compared with a reference value ΔK ₀ stored in 49. When the correction value ΔK is smaller than the negative reference value ΔK _0, it is determined that the air pressure of the tire 26 is abnormally low, and the display device 66 notifies the driver. The relationship between the correction value ΔK and the air pressure change amount ΔP is stored in the ROM 49 in advance, and the air change amount ΔP corresponding to the current correction value ΔK is calculated according to the relationship.

In the wheel speed output section 64, the rim side
The rotation speed v supplied from the rotation speed calculation / correction unit 45 is
Compensation based on the disturbance estimated by the disturbance observer 52
Output after being corrected. As mentioned earlier, the disturbance observer
By the part configured based on the equation (11) of the bar 52,
Estimated disturbance w_2pIs, as shown in equation (8), w_2p=
(-1 / J_B) T _d+ (ΔK / J_B) Θ_RBRepresented by
However, the second term on the right side of this equation is
As described above, it is continuously corrected and changes rapidly.
, So it is negligibly small compared to the first term
No. Therefore, in the wheel speed output unit 64, the disturbance
The portion of the reservoir 52 configured based on the expression (11)
Disturbance w estimated_2pIs (-1 / J_B) T_dIs
Assuming that the rotation speed v is corrected.

Specifically, the disturbance w _2p = (− 1 / J _B ) T
_The disturbance torque T _d is obtained by multiplying _d by −J _B , and (2)
The estimated angular velocity ω _Rp of the rim side portion 28 attributable only to the disturbance torque T _d is obtained by the equation 7). ω _Rp (s) = ｛[D] (s [I] − [E] ⁻¹ [F]｝ T _d (s) (27) where [I]: identity matrix s: Laplace operator ω _Rp (s): Laplace-transformed value of estimated angular velocity ω _Rp T _d (s): Laplace-transformed value of disturbance torque T _d [D], [E], and [F] are expressed by the following equations, respectively. Vectors and matrices.

[0081]

[Equation 11]

[0082] The angular velocity estimate omega _Rp is the disturbance of the rotational speed v of the wheel 14, since a component due to disturbance applied from the road surface to the wheel 14, the wheel the angular velocity estimate omega _Rp 14
The rotational speed v supplied from the rim side rotational speed calculation / correction unit 45 is corrected by the value converted to the peripheral speed of the rim, and noise of the rotational speed v due to disturbance from the road surface is removed.

[0083]

As described above, the first aspect of the present invention provides
Detects the rotation of the rotating body, generates a periodic pulse signal corresponding to the rotating speed of the rotating body, calculates the rotating speed based on the cycle of the pulse signal, is determined in advance and stored in a plurality of registers. A rotation speed that reads a correction value having a predetermined relationship with the rotation speed calculated this time from among a plurality of correction values that are present, and determines a rotation speed that eliminates the periodic disturbance with respect to the rotating body by calculation with the rotation speed calculated this time. In the detection device, edge counting means for counting the number of edges of the pulse signal in a predetermined time period, and an integer part of a value obtained by dividing an addition value of the counted edge number and the carrier in the previous predetermined time period by a predetermined number. Derived number calculating means for calculating as a derived number corresponding to the number of storages of the rotational speed in the current predetermined time period, and the number of edges counted this time
And the added value of the carrier in the previous predetermined time period,
A carrier calculating means for calculating a carrier by subtracting the derived number and the predetermined number of multiplied values, and reading a rotational speed corresponding to the derived number from the plurality of registers to remove a periodic disturbance; Is used to update the rotational speed of the register from which the above-mentioned reading has been performed.

For this reason, it is sufficient to prepare as many registers as the number obtained by dividing the number of edges obtained by one rotation of the rotating body by a predetermined number, and the number of registers can be greatly reduced from the number of registers required conventionally. Therefore, the memory capacity can be significantly reduced.

[Brief description of the drawings]

FIG. 1 is a functional block diagram of the present invention.

FIG. 2 is a configuration block diagram of the present invention.

FIG. 3 is a sectional view showing a part of a wheel.

FIG. 4 is a diagram showing a dynamic model of a wheel.

FIG. 5 is a flowchart of a rotation speed calculation / correction routine.

FIG. 6 is a diagram illustrating frequency characteristics of wheel speed detection values.

FIG. 7 is a transfer function block diagram of a disturbance elimination filter.

FIG. 8 is a diagram illustrating frequency characteristics of a disturbance elimination filter.

FIG. 9 is a diagram conceptually showing a configuration of a memory for one rotation and a method of using the memory.

FIG. 10 is a diagram for explaining the relationship between the contents stored in the memory for one rotation and the sampling period.

FIG. 11 is a diagram for explaining the relationship between the contents stored in the memory for one rotation and the sampling period.

FIG. 12 is a graph conceptually showing the principle of periodic disturbance removal.

FIG. 13 is a flowchart of a disturbance removal routine.

FIG. 14 is a diagram for explaining disturbance removal processing.

FIG. 15 is a graph for explaining a reduction in memory capacity in periodic disturbance elimination.

FIG. 16 is a diagram for describing approximation of disturbance dynamics.

FIG. 17 is a block diagram showing a disturbance observer.

FIG. 18 is a flowchart of a correlation calculation routine for acquiring a change in spring constant.

[Explanation of symbols]

 Reference Signs List 10 rotor 12 electromagnetic pickup 14 wheel (wheel with tire) 20 computer 24 wheel 26 tire 28 rim side 30 belt side 32 torsion spring 47 computer 52 disturbance observer

Continuation of front page (58) Field surveyed (Int. Cl. ⁷ , DB name) G01P 3/42-3/489 G01D 5/245

Claims (1)

(57) [Claims] 1. A method for detecting a rotation of a rotating body, generating a periodic pulse signal corresponding to a rotating speed of the rotating body, calculating a rotating speed based on a cycle of the pulse signal, and determining a plurality of rotation speeds determined in advance. A rotation speed obtained by reading a correction value having a predetermined relationship with the rotation speed calculated this time from a plurality of correction values stored in the register and removing a periodic disturbance with respect to the rotating body by calculation with the rotation speed calculated this time. An edge counting means for counting the number of edges of the pulse signal in a predetermined time period, and an addition value of the counted edge number and a carrier in a previous predetermined time period divided by a predetermined number . and deriving the number calculating means for calculating a derived number to the corresponding integer part of the value to store the number of the rotational speed in a predetermined time period now <br/> times, this time the counted edge Contact a predetermined time period of the last time the
And a carrier calculating means for calculating the carrier by subtracting the derived number and the multiplied value of the predetermined number from the added value of the carrier to be read. A rotation speed detecting device, which is used for determining a rotation speed from which a target disturbance has been removed, and updates the rotation speed of the register from which the above-mentioned reading has been performed.