JP2003287546A - Method and apparatus for detecting speed of rotator - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To solve the problem that occurs in the case of detecting the speed of a wheel with a mean pulse period of one rotation of the wheel as a reference. <P>SOLUTION: By correcting the pulse period of pulse signals being continuously generated according to the rotation of the wheel, the speed of the wheel is detected. It is assumed that a speed change between two adjacent sampling timings is small, and a correction coefficient of the pulse period that minimizes error between an actual rotational speed and a detected speed is computed. The pulse period is corrected by the computed correction coefficient to detect the rotational speed of the rotator. <P>COPYRIGHT: (C)2004,JPO

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method and apparatus for detecting the speed of a rotating body, and particularly to a pulse signal continuously generated in response to the rotation of the rotating body to be measured. The present invention relates to a method and apparatus for detecting a speed of a rotating body, which detects a rotation speed by correcting a detection error based on a non-standardizing factor using a correction coefficient.

[0002]

2. Description of the Related Art In-vehicle control systems, such as ABS,
The wheel speed information necessary for TRC, VSC, etc. is obtained by processing the output of the wheel speed sensor provided on the wheel rotation shaft. This wheel speed sensor is composed of a rotating part (signal rotor) having a plurality of (normally 48) teeth formed on the circumference thereof, and a detecting part composed of a pickup coil, a magnetic sensor and the like for detecting passage of the teeth. A pulse period of a pulse signal formed of electrical pulses continuously generated when the teeth of the rotating portion pass over the detecting portion is counted, and the wheel speed, which is the rotational speed information, is detected from the reciprocal thereof.

Further, although a vibration component due to the uniformity of the tire is also mixed in the wheel speed, this error can be regarded equivalently as a pitch error. Therefore, the following will be mainly treated as correction of the sensor pitch error, and the wheel speed will be described below. Will be described. In order to simplify the explanation below,
Consideration will be given to detection by so-called one-edge calculation that uses only the rising or falling of the pulse of the wheel speed sensor. The double edge calculation and the intermediate edge calculation using both the rising edge or the falling edge or the central time points thereof are the same as described below.

If the period from the rise or fall of a pulse to the rise or fall of the next pulse is t seconds, the wheel speed ω can be obtained by the following equation.

[0005]

[Equation 1]

The expression (1) is an expression in the case where the gear pitch of the wheel speed sensor is uniform and the pitch error is not taken into consideration.
With an actual sensor, a manufacturing error occurs in the tooth length (tooth pitch), so even if the wheel speed is constant, the wheel speed will have a frequency that is an integer multiple of the wheel rotation speed due to the pitch error. Appears and adversely affects various control and estimation systems.

Conventionally used pitch error correction
The method is the average pulse period t when the wheel makes one rotation._aTo
This average pulse period t is calculated by the following equation (2)._aAnd for each tooth
Cycle t _n(N = 1, 2, 3, ..., 48)
Is a constant gain k_aTo feed back the correction coefficient
It is what you want.

[0008]

[Equation 2]

That is, for the average pulse period t _a ,
The correction coefficient W _n of each tooth is obtained by iterative calculation of the following equation (3), the period t _{n of} each tooth is corrected as in the following equation (4), and the correction period t _n ′ is obtained.

[0010]

[Equation 3]

[0011]

[Equation 4]

In this prior art, it has been confirmed that the characteristics become discontinuous during acceleration / deceleration, so that 1 of each pulse period is used.
With respect to the deviation (t _on −t _n ) from the period t _on before rotation,

[0013]

[Equation 5]

The correction term Δt for the speed change represented by
A method of correcting the average pulse period as shown in the following formula (6) has also been proposed.

[0015]

[Equation 6]

However, the logic of the conventional pitch error correction method has the following problems because the average pulse cycle of one rotation of the wheel is used as a reference.

Since a learning calculation is required for each pulse, and most of the calculation processing is spent on the calculation of the wheel speed with the frequent occurrence of pulses at a high speed, the calculation load becomes large. Further, the learning calculation for each pulse has a problem that the calculation frequency changes depending on the speed, it is difficult to achieve consistency with other control systems operating in a constant cycle, and the control system design is difficult to handle. Further, since it takes a long time to converge the learning calculation by the repetition, there is a problem that it cannot follow a rapid uniformity change such as a vehicle speed change. Further, there is a problem that the wheel speed becomes discontinuous when acceleration / deceleration occurs.

The present invention has been made in order to solve the above problems, and it is possible to correct the rotation speed such as the wheel speed at a predetermined cycle smaller than one rotation of the rotating body to be measured such as the wheel. Accordingly, it is an object of the present invention to provide a rotation speed / speed correction method and device that solve the problems that occur when the average pulse cycle of one rotation of the wheel is used as a reference.

Further, according to the present invention, by utilizing the least squares method, the time series near the present time is calculated to improve the followability to the acceleration / deceleration of the rotating body due to the acceleration / deceleration of the vehicle, etc., and it is relatively fast. An object of the present invention is to provide a rotation speed detection method and device having convergence.

[0020]

In order to achieve the above object, the method for detecting the speed of a rotating body according to the present invention provides a pulse period of a pulse signal consisting of pulses continuously generated in accordance with the rotation of the rotating body to be measured. , Or a speed detecting method of a rotating body for detecting a rotating speed of the rotating body on the basis of a pulse generation frequency within a predetermined time shorter than one rotation of the rotating body, wherein a plurality of sample timings located at approaching points are included. Assuming that the change in speed is small, the correction coefficient of the pulse cycle or pulse generation frequency that minimizes the error between the true rotation speed and the detected speed is calculated, and the pulse cycle or the pulse generation is calculated using the calculated correction coefficient. It is characterized in that the frequency is corrected to detect the rotation speed of the rotating body.

The speed detecting device for a rotating body according to the present invention is
A pulse output that outputs a pulse signal composed of a rotating part having a plurality of teeth formed around the rotating part that rotates according to the rotation of the rotating body to be measured, and a pulse signal that is continuously generated according to the passage of the teeth of the rotating part And the pulse frequency of the pulse signal or the pulse generation frequency within a predetermined time shorter than one rotation of the rotating body, it is assumed that the velocity change is small within the plurality of sample timings located at the close time points. , Calculating a correction coefficient for the pulse cycle or pulse generation frequency that minimizes the error between the true rotation speed and the detected speed, and correcting the pulse cycle or the pulse generation frequency with the calculated correction coefficient to rotate the rotating body And a detection unit for detecting the.

In a plurality of sample timings located at close time points, two adjacent sample timings are separated from each other by a predetermined sample timing (for example, one sample timing to several sample timings). It may be within the sample timing, or 3
The above sample timing may be used.

Since the rotation speed is proportional to the reciprocal of the pulse cycle or the pulse generation frequency within a predetermined time, the rotation speed can be detected based on the pulse cycle or the pulse generation frequency within the predetermined time.

In the above invention, based on the pulse period of the pulse signal consisting of pulses continuously generated according to the rotation of the rotating body to be measured, or the pulse generation frequency within a predetermined time shorter than one rotation of the rotating body, When detecting the rotation speed of the rotating body, assuming that the speed change within a plurality of sample timings located at close time points is small, a pulse period or a pulse cycle that minimizes the error between the true rotation speed and the detected speed. The correction coefficient of the pulse generation frequency is calculated, and the pulse cycle or the pulse generation frequency is corrected by the calculated correction coefficient to detect the rotation speed of the rotating body.

In the present invention, the speed of the rotating body is detected based on the pulse period shorter than the one rotation of the rotating body or the pulse generation frequency within a predetermined time shorter than the one rotation of the rotating body. The problem that occurs when the average pulse period of one rotation of the rotating body is used as a reference can be eliminated.

In the present invention, the sum of the periods of the pulses generated in each of the sample timings and the sum of the numbers of the pulses generated in each of the sample timings are obtained, and the sum of the periods and the sum of the numbers of the pulses are obtained. It is possible to calculate the correction coefficient of the pulse period or the pulse generation frequency that minimizes the squared error between the true rotation speed and the detected speed by using and.

By using the least squares method in this way, it is possible to perform a calculation on a time series in the vicinity of the present time to improve the followability to the acceleration / deceleration of the rotating body and detect the rotation speed with a relatively fast convergence. You can

The above-mentioned correction coefficient is the ratio of the cycle of each pulse to the cycle of one rotation when the rotating body rotates at a constant speed, and the cycle of one rotation when the rotating body rotates at a constant speed. Each of the error ratios of the period of each pulse with respect to the average pulse period divided by the number of pulses generated in one rotation, and each of the other for the predetermined pulse period for the period of one rotation when the rotating body rotates at a constant speed. Any of each of the pulse period ratios can be used.

[0029]

BEST MODE FOR CARRYING OUT THE INVENTION Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. In the present embodiment, a signal rotor 10 as a rotating part having a plurality of teeth formed around a rotating body that rotates in accordance with the rotation of a rotating body to be measured, and pulses are continuously generated in response to passage of the teeth of the signal rotor 10. A pulse generator 12 configured by a pick-up or the like, and
An arithmetic circuit configured by a microcomputer or the like that detects the rotation speed of the rotating body to be measured based on the pulse cycle within a predetermined sample timing of a pulse signal formed of pulses that are continuously generated according to the rotation of the signal rotor 10. It is composed of 14. The predetermined sample timing is set to be shorter than one rotation of the rotating body, that is, the pulse signal due to the passage of all the teeth of the signal rotor 10 is not included in the predetermined sample timing.

Hereinafter, a case where the present invention is applied to a wheel speed sensor for detecting a wheel rotation speed (wheel speed) as a rotation speed of a rotating body to be measured will be described as an example. In the following,
Although the number of teeth of the signal rotor is described as 48, the number of teeth is not limited to 48 and may be a predetermined number.

Many on-vehicle control systems operate at a sampling period of about 6 [ms], and a wheel speed sensor having 48 teeth has a low speed of about 10 [km / h] within one operating period. 0 to 1 pulse, and 4 to 5 pulses are generated at a high speed of 100 [km / h].

Now, assuming the pulse generation timing of FIG. 1, the number of pulses in the predetermined sample timing, that is, the number of pulses in the interval based on the sample timing i-1 at the sample timing (sampling time) i is N _i , The pulse period is the total sum T _{i of the} N _i pulses. Although the example in which the pulse generation timing and the sample timing are not synchronized has been described in the drawing, the pulse generation timing (pulse rising edge or pulse falling edge) may be used as the sample timing in synchronization with the pulse generation timing. In this case, there will be only one pulse within each sample timing.

At this time, each sensor pitch when the wheel rotates at a constant speed is k _j (j = 1, 2, 3,
, 48, Σk _j = 1)
Since this ratio becomes a correction coefficient for each tooth, the wheel speed correction value ω can be obtained by the following equation (7). That is,
When the ratio of the cycle of each pulse to the cycle of one rotation when the wheel rotates at a constant speed is used as the correction coefficient, the wheel speed correction value ω can be obtained by the following equation (7).

[0034]

[Equation 7]

Here, s _i is the number of the first tooth of the pulse within the sample timing at time _i , and s _i + N _i -1 is the number of the last tooth of the pulse within the sample timing at time i. There is.

A formula similar to the above formula (7) is calculated by using the adjacent i-
If set within the sample timing at one time point, the following equation (8) is obtained.

[0037]

[Equation 8]

In the conventional correction method, the error of each tooth is corrected on the assumption that the rotation speed is constant during one rotation based on one rotation of the wheel, but in the present embodiment, the adjacent sample timings are corrected. The correction is made assuming that the velocity does not change significantly. Here, as shown in the following equation (9),
It is assumed that the wheel speed correction values ω in equations (7) and (8) are equal. In the above description, it is assumed that the wheel speed correction values ω are equal, but it is also possible to assume that the absolute value of the difference between the wheel speed correction values ω in the expressions (7) and (8) is equal to a minute value.

[0039]

[Equation 9]

By modifying the equation (9), the following equation (10) is obtained.

[0041]

[Equation 10]

Between s _i , s _i-1 and N _i-1 , the following (1
Since there is a relation of formula (1), formula (10) has a tooth number of s.
_This indicates that the sum of the weights of the correction coefficients from _i-1 to s _i + N _i-1 becomes zero.

[0043]

[Equation 11]

The above equation (10) is obtained for each sample timing (sampling period) and summarized in a matrix form as shown in the following equation (12).

[0045]

[Equation 12]

Since the sum of the correction coefficients for one round is 1, the following expression (14) is obtained by combining the conditional expression of the following expression (13) with the above expression (12).

[0047]

[Equation 13]

[0048]

[Equation 14]

This equation (14) is

[0050]

[Equation 15]

Since it has the form of, the least squares method is applied.
can do. Therefore, A on both sides ^TMultiply by (1
Detected as true wheel speed by transforming as in equation 6)
Of the signal rotor that minimizes the squared error from the wheel speed
Pitch of each tooth, that is, pulse cycle correction factor k_jGot
You can

[0052]

[Equation 16]

Further, A ^T and B ^T can be expressed by the following equations.

[0054]

[Equation 17]

However,

[0056]

[Equation 18]

It is A ^T A is a symmetric matrix, but has all elements because all of the conditional expressions are 1.

Further, since H can be expressed by the following equation (19),

[0059]

[Formula 19]

When summarized in the form of sequential calculation,

[0061]

[Equation 20]

From the above,

[0063]

[Equation 21]

As a correction coefficient, the corrected wheel speed can be calculated by substituting the calculated correction coefficient into the above equation (7). However, ρ in the equation (20) is a forgetting factor.

The results of verifying the above-described present embodiment by simulation will be described below. The set pitches of the wheel speed sensor are as shown in Table 1, and it is assumed that a maximum pitch error of about 10% exists. Assume that the pulse interval is counted at a clock cycle of 4 [MHz],
Further, the sampling cycle of calculation is set to 6 [ms].

[0066]

[Table 1]

The simulation results are shown in FIGS. The forgetting factor in this embodiment is 0.999.

FIG. 2 is a simulation result when the wheel speed is constant at 50 [rad / s]. (A) is a set true value, and (B) is a simple value obtained by assuming that tooth pitches are at equal intervals. Calculated values, (C) is a correction value by the conventional method, and (D) is a correction value by the least squares method of the present embodiment. It can be seen from the simple calculation that band-shaped vibration appears due to the pitch error. In the conventional method, the error gradually decreases with the passage of time, and converges to a true value in the vicinity of 6 seconds.

On the other hand, in the present embodiment, 1.5
It can be seen that it converges to the true value rapidly near the second. In the present embodiment, it is necessary to obtain the inverse matrix when calculating the equation (21), but since the matrix has few components and the inverse matrix cannot be obtained at the initial stage of calculation, a value equal to the simple calculation value is output. is doing. It takes about 1.5 seconds to obtain a sufficient matrix component to obtain the inverse matrix, but if the inverse matrix is obtained, a value close to the true value is obtained at that time. In this case, the wheel 10
Sufficient matrix components are obtained to solve the inverse matrix by rotating. This means that the pitch error compensation can be derived in a time series of about 1.5 seconds near the current point, and although it is related to speed as well, the followability to uniformity variation is improved as compared with the conventional method. Can be expected.

Similarly, also in FIG. 3 when acceleration / deceleration is performed, the present embodiment shows sufficient convergence, and correction can be performed without problems in a normal running state. It is presumed that the reason for this is that the effect of acceleration / deceleration is reduced because the error between adjacent sample points is minimized rather than being corrected for each rotation of the wheel.

Although the above-mentioned acceleration / deceleration correction is performed in the conventional method, the convergence speed is slightly slower than the constant speed.

FIG. 4 shows the calculation result of the wheel speed obtained from the wheel speed pulse when the vehicle is running at a constant speed of approximately 70 [rad / s]. (A) is a simple calculation value obtained assuming that the tooth pitches are evenly spaced, (B) is a correction value by the conventional method, and (C) is a correction value by the least squares method of the present embodiment. The pitch error of the actual vehicle is smaller than the value assumed in the simulation,
Even in the simple calculation value, no large band-like vibration was observed as in the simulation, and the three characteristics had almost the same values.

FIG. 5 shows the FFT for the obtained wheel speed.
It is the frequency characteristic of the amplitude calculated. At the wheel rotation speed of 70 [rad / s], the rotational primary vibration is 11 [Hz].
To some extent, in the simple calculation value without correction, a peak characteristic is seen at a frequency that is an integral multiple of this value. In estimation systems such as tire pressure estimation and road surface μ maximum value estimation based on wheel vibration phenomena, we focus on the frequency characteristics near 40 [Hz], and the effect of vibration due to tire pitch error is large, and removal of this component Is inevitable. On the other hand, it can be seen that the rotational first-order component is successfully removed by both the conventional method and the proposed method. A comparison between the conventional method and the proposed method shows that the conventional method has a characteristic that the vicinity of the frequency component to be removed drops too much, and the characteristics other than the component to be removed also change, although they are slight. Further, with respect to the high frequency component of 50 [Hz] or more, it seems that the amplitude value slightly increases in the conventional method.

In the above, the present embodiment for correcting the pitch error of the wheel speed sensor at an arbitrary cycle by the least square method has been described, and its effectiveness has been confirmed by simulation. However, the present invention uses the least square method. However, the method other than the least square method may be used as long as the correction coefficient is calculated so as to minimize the error between the true wheel speed and the detected wheel speed.

The present embodiment does not require the learning calculation for each pulse generation as in the conventional method, but only the simple period and the number of pulses are counted for each pulse generation, and the compensation calculation is performed in a relatively long period. I am doing it. Also, it has been confirmed by simulation that almost a true value is instantaneously obtained when a sufficient time has passed for calculating the inverse matrix, and it is expected that the followability with respect to the uniformity variation can be expected. It was found from the frequency characteristics of the correction signal that the cyclic error with respect to the first-order rotation can be properly corrected.

In the present embodiment, there is a problem that 48th order matrix calculation is required. However, since the integer calculation in the ECU is improved or the correction coefficient of all the teeth is not always required for the calculation of the wheel speed, a few (100 [km / h] required to determine the current wheel speed are required. ] 4-5 when driving
This problem can be dealt with by using a method of calculating only the coefficient) or a method that does not use the inverse matrix calculation such as the online least square method.

In the above, each of the pitch ratio of each tooth to one rotation of the wheel, that is, each of the ratio of the cycle of each pulse to the cycle of one rotation when the wheel rotates at a constant speed is used as the correction coefficient. You may make it use the correction coefficient. Alternatively, each of the ratios of the other pulse cycle to the predetermined pulse cycle with respect to the cycle of one rotation when the wheel rotates at a constant speed may be used as the correction coefficient.

First, each deviation from the average value of each pitch, that is, each pulse with respect to the average pulse cycle obtained by dividing the cycle of one rotation when the wheel rotates at a constant speed by the number of pulses generated in one rotation Each of the error ratios of the periods is used as a correction coefficient.

This uses the deviation W _i from the average value of each pitch as a parameter, instead of using the conventional reciprocal deviation. The value of the wheel speed ω is

[0080]

[Equation 22]

And

[0082]

[Equation 23]

And the conditional expression

[0084]

[Equation 24]

Calculate from Similarly, A ^T A is a symmetric matrix, and since all the conditional expressions are 1, it has all the elements.

Secondly, each of the ratios to arbitrary teeth, that is, the ratio of each other pulse period to a predetermined pulse period to one rotation period when the wheel rotates at a constant speed is used as a correction coefficient. It is a thing.

For example, the ratio l to the tooth pitch where i = 1
_i is a parameter, and the correction coefficient k _{i in} equation (13) is
When,

[0088]

[Equation 25]

Since there is a relation of

[0090]

[Equation 26]

Are of the same type. The conditional expression is

[0092]

[Equation 27]

It is In this case, similarly, A ^T A is a symmetric matrix, but since the conditional expression has only one element, it becomes a band matrix gathered near the diagonal.

In the conventional method of deriving the average of one rotation of the wheel and repeatedly obtaining the deviation from the average, if a large change in the rotation speed occurs during one rotation of the wheel, a large change in the average value occurs. The correction coefficient has a discontinuous stepwise change. On the other hand, in the present embodiment, a solution that minimizes the squared error is sought on the assumption that there is little change in the rotation speed near the corresponding sample point, so a relatively large change in rotation speed occurs. However, continuous values can be obtained accurately.

Although the case where the wheel speed is detected based on the pulse cycle has been described above, the wheel speed may be detected using the pulse generation frequency within a predetermined time shorter than the time for one rotation of the wheel. Good. Further, in the above, the example in which the correction coefficient of the pulse period that minimizes the square of the error between the true rotation speed and the detected speed is calculated assuming that the speed change is small between two adjacent sample timings has been described. However, the correction coefficient may be calculated on the assumption that the velocity change is small within the two sample timings that are apart from each other within the predetermined sample timing.

[0096]

As described above, according to the present invention, the speed of the rotating body is detected based on the pulse period or the pulse generation frequency within a time shorter than one rotation time of the rotating body. It is possible to solve the problem that occurs when the average pulse period of one rotation of the body is used as a reference.

If the least squares method is used in the present invention, the time series near the present time can be calculated to improve the followability to the acceleration / deceleration of the rotating body and detect the rotation speed with a relatively fast convergence. The effect is obtained.

[Brief description of drawings]

FIG. 1 is a block diagram of an embodiment of the present invention.

FIG. 2 is a simulation result when the wheel speed is constant at 50 [rad / s], (A) is a set true value,
(B) is a simple calculation value obtained assuming that the tooth pitches are evenly spaced, (C) is a correction value by the conventional method, and (D) is a correction value by the least squares method of the present embodiment.

FIG. 3 is a diagram similar to FIG. 2 when acceleration / deceleration is performed.

FIG. 4 is a calculation result of a wheel speed obtained from a wheel speed pulse when the vehicle travels at a constant speed of approximately 70 [rad / s],
(A) is a simple calculation value obtained assuming that the tooth pitches are evenly spaced, (B) is a correction value by the conventional method, and (C) is a correction value by the least squares method of the present embodiment.

FIG. 5 is a diagram showing the frequency characteristics of the amplitude obtained by FFT calculation with respect to the obtained wheel speed, (A) is a simple calculation value obtained assuming that the tooth pitches are at equal intervals, and (B) is a diagram. Is a correction value by the conventional method, and (C) is a correction value by the least squares method of the present embodiment.

[Explanation of symbols]

10 signal rotor 12 pulse generator 14 Arithmetic circuit

   ─────────────────────────────────────────────────── ─── Continued front page    (72) Inventor Katsuhiro Asano             Aichi Prefecture Nagachite Town Aichi District             Ground 1 Toyota Central Research Institute Co., Ltd. F term (reference) 3D046 BB28 BB29 HH36

Claims (6)

[Claims] 1. Based on the pulse period of a pulse signal composed of pulses continuously generated according to the rotation of a rotating body to be measured, or the pulse generation frequency within a predetermined time shorter than one rotation of the rotating body, A method for detecting the rotational speed of the rotating body, which is a method for detecting the rotational speed of the rotating body, wherein a true rotational speed and a detected speed are assumed to be small on the assumption that a speed change within a plurality of sample timings located at approaching points is small. A method for detecting a speed of a rotating body, which calculates a correction coefficient of a pulse cycle or a pulse occurrence frequency that minimizes an error, and corrects the pulse cycle or the pulse occurrence frequency with the calculated correction coefficient to detect a rotation speed of the rotating body. 2. The square of the true rotation speed and the detected speed is obtained by calculating the sum of the periods of the pulses generated in each of the sample timings and the sum of the number of pulses generated in each of the sample timings. A correction coefficient for the pulse period or pulse generation frequency that minimizes the error is calculated.
A method for detecting the speed of a rotating body according to the description. 3. The ratio of the period of each pulse to the cycle of one rotation when the rotary body rotates at a constant speed, and the cycle of one rotation when the rotary body rotates at a constant speed is generated for one rotation. Error ratio of the period of each pulse to the average pulse period divided by the number of pulses to be generated, and the ratio of each other pulse period to a predetermined pulse period for one rotation period when the rotating body rotates at a constant speed. The method for detecting a speed of a rotating body according to claim 1 or 2, wherein any one of the above is used as a correction coefficient. 4. A pulse signal composed of a rotating portion having a plurality of teeth formed around the rotating body that rotates in response to rotation of a rotating body to be measured, and a pulse continuously generated in response to passage of the teeth of the rotating portion. And a pulse output section for outputting a pulse frequency of the pulse signal or a pulse generation frequency within a predetermined time shorter than one rotation of the rotating body, the speed change within a plurality of sample timings located at approaching points. Assuming a small value, calculate a correction coefficient for the pulse cycle or pulse generation frequency that minimizes the error between the true rotation speed and the detected speed, and correct the pulse cycle or the pulse generation frequency with the calculated correction coefficient. A speed detecting device for a rotating body, comprising: a detecting unit that detects a rotating speed of the rotating body. 5. The square of the true rotation speed and the detected speed is obtained by calculating the sum of the periods of the pulses generated in each of the sample timings and the sum of the number of pulses generated in each of the sample timings. A correction coefficient for the pulse period or pulse generation frequency that minimizes the error is calculated.
A speed detecting device for the rotating body described. 6. A cycle of one rotation when the rotating body rotates at a constant speed, and a cycle of one rotation when the rotating body rotates at a constant speed, are generated in one rotation. Error ratio of the period of each pulse to the average pulse period divided by the number of pulses to be generated, and the ratio of each other pulse period to a predetermined pulse period for one rotation period when the rotating body rotates at a constant speed. The speed detecting device for a rotating body according to claim 4 or claim 5, wherein any one of the above is used as a correction coefficient.