JPH1038576A - Azimuth sensor - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an azimuth sensor wherein the effect of disturbance in measurement can be separated and self-examination based on angular drift evaluation is possible. SOLUTION: A sensing group 2 comprising an optical fiber is stood on a rotative table 1 which rotates with a gravity axis as the center, and the rotative table 1 is rotated by a specified angle θ from a reference axis representing its reference angle, and from an angular speed measurement value of the sensing group 2 based on globe's rotation measured then, the azimuth of the reference axis is detected. To the angular speed measurement value at multiple angles of the rotative table, Fouurier transformation is applied, so that a spectral distribution is calculated, and the phase of basic cycle component of the spectrum is taken as azimuth of the reference axis.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an azimuth sensor using an optical fiber gyro, and more particularly to an azimuth sensor capable of separating the influence of disturbance during measurement and performing self-diagnosis by evaluating angular velocity drift. is there.

[0002]

2. Description of the Related Art As shown in FIG. 1, a azimuth sensor using an optical fiber gyro has a sensing loop 2 made of an optical fiber on a turntable 1 which rotates around a gravity axis.
The direction of the sensing loop 2 is changed by rotating the turntable 1 by a predetermined angle from a reference axis (azimuth sensor reference axis) indicating the reference angle of the azimuth sensor. At this time, the sensing loop is set based on the rotation of the earth. 2 is measured, and the azimuth of the azimuth sensor reference axis is detected from the measured value (earth rotation angular velocity measured value).

Here, if the azimuth of the reference axis of the azimuth sensor with respect to true north is Φ, the maximum value of the measured earth rotation angular velocity measured in one round of the sensing loop is A, and the angular velocity bias of the sensing loop is B, The measured value of earth rotation angular velocity Ω is expressed as follows: Ω = A · sin Φ + B (1)

After the start of the azimuth sensor, the azimuth sensor reference axis is set to the azimuth that is approximately north, and then the sensing loop is directed roughly to three directions of north, east or west, and south, and the measured value of the earth rotation angular velocity Ω in each direction. _N, Ω _E, to measure the Ω _S.

Ω _N = A · sin Φ _N + B (2) Ω _E = A · sin (Φ _N +90) + B = A · cos Φ _N + B (3) Ω _S = A · sin (Φ _N +180) + B = −A Since sin Φ _N + B (4), B = (Ω _N + Ω _S ) / 2 (5) Φ _N = tan ⁻¹ {(Ω _N −B) / (Ω _E −B)} (6) , The azimuth φ _N of the azimuth sensor reference axis with respect to true north is determined.

In addition, there is a method of obtaining the azimuth φ _{N of the} reference axis of the azimuth sensor by least-squares approximation of the measured value of the earth rotation angular velocity Ω according to the equation (1).

[0007]

In the prior art method for solving a trigonometric function, noise caused by disturbance applied to an azimuth sensor at the time of measuring angular velocity can be separated from a fundamental period component (described later).
It cannot be detected only.

In the method using the least square approximation, although the fundamental period component can be detected, the magnitude of the disturbance at the time of measuring the angular velocity cannot be calculated, so that the reliability of the detected azimuth cannot be evaluated.

SUMMARY OF THE INVENTION It is an object of the present invention to solve the above-mentioned problems and to provide an azimuth sensor capable of separating the influence of disturbance during measurement and performing self-diagnosis by evaluating angular velocity drift.

[0010]

SUMMARY OF THE INVENTION In order to achieve the above object, the present invention sets up a sensing loop made of an optical fiber on a rotary table rotating about a gravity axis, and moves the rotary table from a reference axis indicating its reference angle. Rotate a predetermined angle,
A directional sensor that detects the azimuth of the reference axis from the measured angular velocity of the sensing loop based on the rotation of the earth measured at that time, calculates the spectral distribution by performing a Fourier transform on the measured angular velocity at a plurality of angles of the turntable. Then, the phase of the fundamental period component of the spectrum is used as the azimuth of the reference axis.

[0011] The rotary table is continuously rotated by a predetermined angle, and each time, the angular velocity measurement value measured during the past one revolution is subjected to the Fourier transform, and the obtained azimuth is moving averaged for a predetermined number of rotations. You may.

2 for the fundamental period component of the above spectrum
The ratio of the double frequency period component is obtained, and when this ratio exceeds a predetermined value, it is determined that disturbance has occurred during the measurement, and the azimuth obtained at that time may be excluded from the moving average sample.

An angular velocity drift of the sensing loop is obtained from a change in the DC component of the spectrum. When the angular velocity drift exceeds a predetermined value, the azimuth obtained at that time is excluded from the moving average sample or it is determined that a failure has occurred. You may.

[0014]

An embodiment of the present invention will be described below in detail with reference to the accompanying drawings.

As shown in FIG. 1, the azimuth sensor according to the present invention has a sensing table 2 made of an optical fiber on a rotary table 1 which rotates about a gravity axis.
The direction of the sensing loop 2 can be changed by rotating the turntable 1 by an arbitrary angle from the reference axis of the direction sensor, and the measured value of the earth rotation angular velocity can be measured. Reference numeral 3 denotes a circuit board for signal processing and rotation control.

The direction sensor of the present invention has the following features in signal processing.

The measured values of the earth rotation angular velocity in a plurality of directions of the sensing loop are analyzed by Fourier transform. That is, the phase is obtained for the fundamental period component of the spectrum, and this phase is used as the azimuth of the azimuth sensor reference axis. This removes noise in the detection direction due to disturbance during measurement.

The rotary table is continuously rotated by a predetermined angle, for example, 90 °, and the sensing loop is directed in four directions at 90 ° intervals. Only the four angular velocity measurement values measured during this one round are converted to a Fourier transform. Offer. Thus, high-speed measurement is achieved, and the speeding-up reduces the detected azimuth error due to angular velocity drift.

The azimuth obtained each time the rotary table is rotated is moving averaged. Thereby, the resolution of the direction sensor is improved.

The ratio of the double frequency period component to the fundamental period component of the spectrum is obtained, and the magnitude of the disturbance is evaluated based on the ratio. The detection orientation when the disturbance is large is excluded from the moving average sample to improve the reliability.

Similarly, the angular velocity drift of the sensing loop is evaluated. The angular velocity drift is obtained from a change in the DC component of the spectrum. The detection direction when the angular velocity drift is large is excluded from the moving average sample. A self-diagnosis function for determining a failure of the direction sensor based on the angular velocity drift may be provided.

Hereinafter, specific signal processing according to the present embodiment will be described.

First, the rotary table 1 is continuously rotated by a predetermined angle, for example, 90 °, and the sensing loop 2 is turned in four directions at 90 ° intervals, and the earth rotation angular velocity measured values Ω ₀ , Ω ₉₀ , Ω ₁₈₀ , Ω Measure ₂₇₀ . The four measured angular velocities measured during this one round are subjected to Fourier transform. In the present invention, a discrete Fourier transform (DFT) is used, but in order to reduce the operation time, a fast Fourier transform (FF) is used.
T) may be used.

Here, the discrete Fourier transform will be briefly described. The discrete Fourier transform is for converting a discrete sampling value of a signal on a time axis into a discrete intensity spectrum on a frequency axis, and uses a fast Fourier transform. The calculation time can be dramatically reduced. According to the discrete Fourier transform, an integer N points are sampled on a time axis of a periodic signal f (t) having a certain period T, and the value is represented by fn
(N = 0, 1,..., N−1), and a constant Wn = exp
Defining (-j2π / N), the discrete Fourier transform is

[0025]

(Equation 1)

## EQU1 ##

Since fn is defined as a set of N complex numbers, equation (7) is represented as equation (8).

[0028]

(Equation 2)

Re and Im represent a real part and an imaginary part of a variable following the symbol, respectively. Thus F
(K) is a complex number, and its intensity (amplitude) is obtained from the mean square of the real part and the imaginary part. The fast Fourier transform is
Focusing on the periodicity of the signal f (t), the arithmetic processing is efficiently improved.

As an example, FIG. 2 shows an example in which a discrete voltage sample of a signal on the time axis is converted into a discrete intensity spectrum on the frequency axis. The voltage signal shown in FIG. 2A is a periodic signal f (t) having a basic period T. Circles are sampling points. Fig. 2 (b) by discrete Fourier transform
Is obtained. Looking at this spectrum distribution, the spectrum discretely has a DC component F ₀ , a fundamental period component F ₁ having a frequency corresponding to the fundamental period, a double frequency period component F ₂ having twice the frequency thereof, and so on. ing.

Next, in the azimuth sensor of the present invention, the rotation angle θ of the turntable with the azimuth sensor reference axis as the reference angle.
FIG. 3A shows the change of the measured value of the earth rotation angular velocity Ω with respect to the rotation axis θ when the horizontal axis represents the rotation angle θ. Circles are sampling points. Paying attention to this change waveform, it can be seen that the signal is a periodic signal that repeats every 360 ° (2π) of the rotation angle of the turntable. Because of this periodicity, it can be seen that the above discrete Fourier transform can be applied. In this case, FIG.
And the time t correspond to the measured value of the earth rotation angular velocity Ω and the rotation angle θ, respectively. After the conversion, a distribution of the discrete intensity spectrum is obtained on the frequency axis as shown in FIG. The period is 1 / (2π).

In the present embodiment, the sampling points in FIG. 3 are four at 90 ° intervals. The measured value of each point is Ω ₀ , Ω ₁ ,
Ω ₂ and Ω ₃ . These measured values correspond to fn (n = 0, 1, 2, 3) when N = 4 in equation (7). Then, by fast Fourier transform, Fk (k = 0,
1, 2, 3) are obtained. That is,

[0033]

(Equation 3)

## EQU1 ## The matrix representation of the above equation is

[0035]

(Equation 4)

## EQU1 ##

In this manner, the DC component F ₀ , the fundamental cycle component F ₁ , the double frequency cycle component F ₂ , and the triple frequency cycle component F ₃ are obtained. The real term Rn and the imaginary term In of each component are obtained, and the amplitude | Fn | and the phase χn of each component are also obtained.

[0038]

(Equation 5)

The phase χ ₁ of the fundamental period component F ₁ is the azimuth Φ _N of the azimuth sensor reference axis.

Next, the moving average will be described.

Since the rotary table 1 is rotated by 90 °, the sampling points # 1, # 2, # 3, # 4, # 5, # 5 as shown in FIG. 6,
... is obtained. In each case, the azimuth is obtained using four sampling points during the past one round. Then, the first time is # 1 to # 4, the second time is # 2 to # 5, and the third time is # 3 to ##
The orientation is obtained every time by using 6,. In this way, even when the positions of the four points move sequentially, the azimuth Φ _N of the azimuth sensor reference axis is obtained every time. This is because only the sensing loop 2 rotates during this time, and the azimuth sensor reference axis does not significantly rotate. The samples of the azimuth Φ _N sequentially obtained in this manner are averaged by a predetermined number in order. The output direction Φ _N by the moving average is

[0042]

(Equation 6)

Here, m is the number of samples used for the moving average and affects the response frequency of the direction sensor.
The determination may be made according to a request from a host system using the direction sensor. p is the first sample number of the sample to be averaged, and Φk is the sample of the azimuth sequentially obtained. In general, when the random output is averaged, the resolution of the sensor is improved by 1 / (square root m) with respect to the sample number m, and the resolution is improved by moving-averaging the past Φk every m past. And a real-time output direction Φ _N can be obtained.

Next, evaluation of disturbance and evaluation of angular velocity drift will be described.

In the spectrum distribution by the Fourier transform of the equation (9), if the measured earth rotation angular velocity Ω is a pure sine function or cosine function of the rotation angle θ, the amplitude of the DC component F ₀ and the double frequency period component F The amplitude of ₂ goes to zero. That is, there is no bias in the angular velocity measurement of the sensing loop,
In an ideal state with no disturbance, the amplitude of the DC component F _{0 and} the amplitude of the double frequency cycle component F ₂ become zero.

When deviated from the ideal state, the DC component F ₀
The amplitude of the amplitude and double frequency period component F ₂ increases.
The amplitude of the DC component F ₀ corresponds to a bias, and the angular velocity drift can be obtained by calculating the change over time. If the angular velocity drift is large, an error occurs in the calculation of the azimuth, so that it is excluded from the moving average processing. If it is determined from the magnitude of the angular velocity drift that the azimuth sensor has failed, the diagnostic data is output.

A method for obtaining the angular velocity drift will be described. The amplitude | F ₁ | of the fundamental period component F ₁ is equal to the maximum value Ω _MAX = 15 (° / h) × cos の (ψ is the latitude) of the measured earth rotation angular velocity Ω. Therefore, the offset angular velocity can be converted by (F ₀ / F ₁ ) × Ω _MAX . This offset angular velocity is calculated for each rotation, and the angular velocity drift is obtained from the amount of change in the offset angular velocity with respect to each time interval.

On the other hand, the amplitude of the double frequency period component F ₂ is
Since this occurs when the sine function or the cosine function is distorted, the magnitude of the disturbance applied to the direction sensor during measurement can be evaluated. Here, 2 for the fundamental period component F ₁
Determining whether removed from the moving average processing with the value of the amplitude ratio of the double frequency periodic component F _2. That is, the detection direction when the ratio is larger than the predetermined value is excluded from the samples of the moving average, and the moving average is suspended until the next sample is obtained. As a result, the detection direction including noise is discarded.

Next, an example of a signal processing procedure according to the present invention will be described with reference to a flowchart. As shown in FIG. 5, first, align the theta rotation angle of the rotary table in the direction sensor reference axis which is the reference angle (Step 1), to measure the Earth's rotation angular velocity measurements Omega _theta (Step 2). Next, the direction of the sensing loop is changed by rotating the turntable by 90 ° (step 3). After the rotation has stopped, the measured value of the earth rotation angular velocity Ω _θ is measured (step 4). It is checked whether the measured values of four points have been acquired, and if not, the turn table is further rotated by 90 ° (step 5). This is 1
This is performed because the measurement values of four points # 1 to # 4 are required for the second direction detection. From the second time, acquisition of four measurement values has already been completed. If four measurement values have been obtained,
An FFT is performed on the measured values of the four points, and the azimuth detection calculation described in the above embodiment is performed (step 6). Thereafter, if the DC component F ₀ is larger than the predetermined value K ₁ , the rotary table is further rotated by 90 ° (step 7). Here, the detected azimuth is discarded by immediately evaluating the DC component F _{0, that} is, the bias, instead of obtaining and evaluating the angular velocity drift as in the above-described embodiment. Note that an error can be output here. Subsequently, if the double frequency period component F ₂ / basic period component F ₁ is larger than the predetermined value K ₂ , the rotary table is also rotated by 90 ° (step 8). This is a disturbance evaluation. Again, an error can be output. In this way, only a reliable detection direction can be used as a moving average sample.
Output orientation Φ by moving average if required number of samples
_N is obtained (step 9). The output direction Φ _N is displayed or transferred to the host system (step 10). Then, the direction sensor reference axis is corrected to true north by changing the numerical value (step 11). In this method, accurate measurement at four points of north, east, south, and west is performed by accurately pointing the azimuth sensor reference axis to true north, thereby improving azimuth accuracy.

Another embodiment and application of the present invention will be described.

In the present embodiment, one azimuth is obtained by using four sampling points, but is not limited to four points. Considering the angular velocity drift, four points are preferable. 4
When using the more sampling points points, higher order spectral components obtained can be a comparison of the amplitude and the fundamental period components F ₁ to the error determination. However, when fast Fourier transform is used, 2 ⁿ (n is a natural number) sampling points are required.

DC component F when the direction sensor is started
Evaluate the bias level at ₀ and the amplitude ratio of the fundamental period components F _1, it can be used for self-diagnosis of the azimuth sensor.

The self-diagnosis result of the azimuth sensor and the evaluation information of the disturbance are supplied to a host system using the azimuth sensor. Thus, the host system can determine whether to use the output direction of the direction sensor. The upper system uses the above evaluation information as a guide when taking measures to prevent disturbance from being applied to the direction sensor. Thus, stable operation of the direction sensor can be easily realized in the host system.

[0054]

The present invention exhibits the following excellent effects.

(1) By performing the Fourier transform, noise due to disturbance can be largely separated from the detected direction, and the accuracy is improved.

(2) Since the magnitude of the disturbance can be evaluated, the reliability of the detected azimuth is improved.

(3) Since the angular velocity drift can be evaluated,
The reliability of the detection direction is improved.

(4) The self-diagnosis based on the evaluation of the magnitude of the disturbance and the evaluation of the angular velocity drift becomes possible, and the reliability of the azimuth sensor itself is improved.

[Brief description of the drawings]

FIG. 1 is a perspective view of an orientation sensor according to an embodiment of the present invention.

FIG. 2 is a time axis characteristic diagram and a frequency axis characteristic diagram showing a general example of a discrete Fourier transform.

3A and 3B are a rotation angle axis characteristic diagram and a frequency axis characteristic diagram in the direction sensor of the present invention.

FIG. 4 is a diagram showing a waveform of the earth rotation angular velocity measured by the direction sensor of the present invention and sampling points.

FIG. 5 is a flowchart showing an example of a signal processing procedure according to the present invention.

[Explanation of symbols]

 1 Rotary table 2 Sensing loop 3 Circuit board

Claims (4)

[Claims] 1. A sensing loop made of an optical fiber is set up on a rotary table rotating about a gravity axis, and the rotary table is rotated by a predetermined angle from a reference axis indicating its reference angle, based on the rotation of the earth measured at that time. In a direction sensor that detects the direction of the reference axis from the measured angular velocity of the sensing loop, a spectrum distribution is calculated by performing a Fourier transform on the measured angular velocity at a plurality of angles of the turntable, and the spectrum has a fundamental period component. An azimuth sensor wherein the phase is the azimuth of a reference axis. 2. The rotary table is continuously rotated by a predetermined angle, and each time, the angular velocity measurement value measured during the past one revolution is subjected to the Fourier transform, and the obtained azimuth is moved by the predetermined number of rotations. The azimuth sensor according to claim 1, wherein averaging is performed. 3. A ratio of a double frequency period component to a fundamental period component of the spectrum is obtained. When the ratio exceeds a predetermined value, it is determined that disturbance has occurred during measurement, and the obtained direction is 3. The azimuth sensor according to claim 2, wherein the azimuth sensor is excluded from a moving average sample. 4. An angular velocity drift of the sensing loop is obtained from a change in a DC component of the spectrum, and when the angular velocity drift exceeds a predetermined value, the azimuth obtained at that time is excluded from a sample of the moving average or a fault is detected. The azimuth sensor according to claim 2, wherein the determination is made.