JPH0228085B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a method for determining the attitude of an artificial satellite based on stellar identification.

To improve the accuracy of satellite attitude control,
It is thought that it will be necessary to realize a system that determines the attitude of an artificial satellite using a stellar sensor.

By the way, this type of attitude determination method that has been developed conventionally uses the observed value of the stellar direction unit vector and the cataloged star corresponding to the observed star as known, and then uses an attitude estimation filter to find the optimal estimated value of the attitude. , using this as an initial value, the attitude was calculated sequentially based on the inertial sensor data.

However, in the conventional method described above, as is well known, it is difficult to determine how to determine the catalog star corresponding to the observed star (i.e., the method of star identification).
remained a problem. Therefore, it has been difficult to realize a practical system.

Furthermore, when using a filter to obtain the optimal estimate of the attitude, a memory is required to store the satellite observation data, and the amount of calculation is enormous and it takes a long time to obtain the estimation results. It was hot.

This invention relates to a method for determining the attitude of an artificial satellite that will be necessary to realize a high-precision three-axis attitude control satellite that is expected to be developed in the future. Equipped with an attitude determination calculator that is equipped with a function to calculate relative attitude changes based on inertial sensor data, and is configured to prevent system malfunctions due to incorrect star identification The aim is to provide a method for determining the attitude of artificial satellites.

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

FIG. 1 is a diagram showing the concept of the attitude determination method according to the present invention. In the figure, 1 is a star sensor, 2
is an inertial sensor, 3 is a stellar vector calculation device, 4 is a database, 5 is a sub-catalog editing device, 6 is a star identification processing device, 7 is a matching matrix calculation device, 8 is an attitude displacement calculation device, 9 is an attitude initial value update 10 is an attitude calculation device, 11 is a mode changeover switch, and 12 is an attitude determination computer.

In such a configuration, by inputting the stellar coordinates (y _i , z _i ) (where i is the number of the observed star) within the sensor field of view observed by the stellar sensor 1, the stellar vector calculation device 3 calculates the coordinates relative to the satellite axis coordinates. Calculate the unit vector $ _i (t ₀ ) in the stellar direction. The subcatalog editing device 5 inputs the star catalog in the database 4 and edits a subcatalog $ _j ^a (j is the catalog star number). At this time, posture information is required, and in the first processing, for example, a predicted value of the posture set in advance in the database is used.
The identification processing device 6 receives the above $ _i and $ _j ^a as input and determines the catalog star $ ^a _ij corresponding to the observed star $ _i (t). In the first process, the mode changeover switch 11 is set to M ₁ =M ₂ =0, and the matching matrix calculation device 7 inputs the above $ _i and $ ^a _ij to create a matching matrix that provides the relationship between the observed value and the corresponding catalog star. Calculate. The attitude displacement calculation device 8 inputs the angular velocity ω _l (where l = 1, 2, 3) around the satellite axis measured by the inertial sensor 2 and calculates the artificial Calculate the change in attitude of the satellite [〓C _B ]. The attitude initial value updating device 9 receives the output of the matching matrix calculation device 7 and [C _B ] as input, calculates a coordinate transformation matrix corresponding to the matching matrix at time t, and outputs it to the subcatalog editing device 5. This completes the first process and turns on the mode switch 11.
Let M ₁ =M ₂ =1. Star sensor 1 at time t=T
Obtain new star observation values and repeat the above series of processing. At this time, the subcatalog editing device 5 edits the subcatalog corresponding to the matching matrix. The star identification processing device 6 receives the subcatalog corresponding to the matching matrix and the output signal of the star vector calculation device 3 as input, and identifies the observed star. On the other hand, the attitude displacement calculation device 8 calculates the fixed star observation time t.
= Calculate the change in posture from T to the present moment.
The attitude initial value updating device 9 receives the output signals of the star identification processing device 6 and the attitude displacement calculation device 8 as input, and calculates the attitude at time t (for example, t=2T). Using this as an initial value, the attitude calculation device 10 receives the inertial sensor data as input, and sequentially calculates and outputs the attitude. After that, read the star sensor data again and repeat the above operation to update the initial attitude value every 2T hours. In each first process after time t=2T, the configuration is such that the attitude calculated by the attitude calculation device 10 at the fixed star observation time is fed back to the subcatalog editing device 5. The attitude determination computer 12 configured in this manner updates the initial value of the attitude every 2T based on stellar identification, and realizes a system that uses this as the initial value to frequently determine and output the attitude within the 2T time. It is something that makes you Furthermore, since the initial posture value is updated based on the results of identification processing performed multiple times (twice in this embodiment), malfunctions of the system due to identification errors can be prevented.

The details of each device constituting the attitude determination computer 12 will be described below with reference to FIGS. 2, 3, and 4.

FIG. 2 is a diagram showing the relationship between stellar sensor coordinates (X _S , Y _S , Z _S ) and satellite axis coordinates (X _B , Y _B , Z _B ). In this example, the number of installed star sensors 1 is 2.
It is said that

In the figure, α ₁ , δ ₁ , α ₂ , and δ ₂ are the star sensors, respectively, and α ₂ and δ ₂ are the optical axis directions of the star sensor 1, respectively.
X _S (azimuth, elevation). In such a relationship, the coordinate transformation matrices [ _B C _S ] ₁ and [ _B C _S ] ₂ that give the relationship between the star sensor coordinates and the satellite axis coordinates are given by the following equations.

[ _B C _S ] ₁ = A ₁₁ A ₁₂ A ₁₃ A ₂₁ A ₂₂ A ₂₃ A ₃₁ A ₃₂ A ₃₃ (1) Here, A ₁₁ = cosα ₁・cosδ ₁ A ₁₂ = −sinα ₁ A ₁₃ = −cosα ₁・sinδ ₁ A ₂₁ = sinα ₁・cosδ ₁ A ₂₂ = cosα ₁ A ₂₃ = −sinα ₁・sinδ ₁ A ₃₁ = sinδ ₁ A ₃₂ = 0 A ₃₃ = cosδ ₁ [ _B C _S ] = B ₁₁ B ₁₂ B ₁₃ B ₂₁ B ₂₂ B ₂₃ B ₃₁ B ₃₂ B ₃₃ (2) Here, B ₁₁ = −sinα ₂・cosδ ₂ B ₁₂ = cosα ₂ B ₁₃ = sinα ₂・sinδ ₂ B ₂₁ = cosα ₂・cosδ ₂ B ₂₂ =-sinα ₂ B ₂₃ =-cosα ₂・sinδ ₂ B ₃₁ = sinδ ₂ B ₃₂ = 0 B ₃₃ = cosδ ₂ Figure 3 shows the observation concept of the unit vector $ _i (t) in the star direction with respect to the satellite axis coordinate. It is a diagram.

In the figure, y _i and z _i are stellar coordinates within the field of view of the stellar sensor, and are values observed by the stellar sensor 1. α _l and δ _l are the (azimuth angle, elevation angle) of the star sensor number l (l=1, 2) in the optical axis direction.

In this relationship, the stellar vector calculation device 2 calculates the unit vector $ _i (t) in the stellar direction with respect to the satellite axis coordinate using the following equation.

$ _i (t) = [ _B C _S ] _l cosy _i・cosz _i siny _i・cosz _i sinz _i l=1, 2i=1, 2, 3 (3) However, in this example, two star sensors 1 Out of multiple observed values, total 3, i.e. $
₁ (t), $ ₂ (t), and $ ₃ (t) are selected and used in the following processing.

On the other hand, the sub-catalog editing device 5 receives [ _B C _I ] _t from the database 4 and the attitude calculation device 11.
(However, only for the first processing, for example, using [ _B C _I ] ₀ set in advance in the database) as input,
Edit the subcatalog $ _j ^a using the following method.

First, the unit vector in the optical axis direction of the stellar sensor 〓
_sl (t) is calculated using the following formula.

〓 _sl (t)=[ _S C _B ] _l [ _B C _I ] _t [1, 0, 0] ^T , l=1, 2 (4) Next, the stellar catalog ^a in database 4
Based on the following conditional expression, a star $ _j ^a that is predicted to exist within the field of view of the star sensor is selected.

$ _j ^a ∈ [ ^a | | cos ^-1 { ^a・〓 _sl } | <ε ₁ ] (5) However, ε ₁ is set to a constant value taking into account the size of the field of view of the stellar sensor.

The star identification processing device 6 uses $ given by the above equation (3).
The following identification process is performed using _i (t), i=1, 2, 3, and the stellar catalog value $ _j ^a given by equation (5).

First, calculate a ₁ = $ ₁ (t)・$ ₂ (t) a ₂ = $ ₂ (t)・$ ₃ (t) a ₃ = $ ₃ (t)・$ ₁ (t) (6) .

Next, take one star $ ₁ ^a from the subcatalog and use the preset constant ε ₂ for the remaining j−1 stars in the subcatalog to calculate $ ₁ ^a , $ _j−1 ^a − It is tested whether there is even one combination that satisfies a _i >cosε ₂ , i=1, 2, 3 (7). If even one pair exists, that star $ ₁ ^a
is left as an identification candidate. Also, if no pair exists, that star is removed from the subcatalog. Repeat this operation for all stars in the subcatalog. If ε ₂ is set to a sufficiently small value, the set of catalog stars corresponding to a ₁ , a ₂ , a ₃ in the subcatalog $ _1j ^a・$ _2j ^a , $ _2j ^a・$ _3j ^a , $ _3j ^a・$ _1j ^a
remains. Therefore, the catalog value of $ ₁ is $ ^a _1j ,
The catalog value of $ ₂ is $ _2j ^a , and the catalog star of $ ₃ is $
_3j ^a , and the identification is completed. However, depending on the conditions, the identification may not be uniquely determined. Thereafter ^{, two sets of catalog stars ($ 1j a , $ 2j a , $ 3j a ) and (S 1j ′, S 2j a , S 3j a} _{) correspond to} ^the _{observed stars (} ^$ ₁ ^, _$ ² , $ ₃ ). The explanation will proceed based on the assumption that the identification remains as an identification candidate and the identification is not uniquely determined.

The mode changeover switch 11 operates as follows in FIG. 1b.

M = 1, ($) = ($) M = 0, ($) = ($) (8) In the first process, M ₁ = M ₂ = 0, and the matching matrix calculation device 7 performs the above stellar identification process. Device 6
Output signals of ($ ₁ , $ ₂ , $ ₃ ), ($ _1j ^a , $ _2j ^a , $ _3j
^a ), ($ _1j ′, $ _2j ^a , $ _3j ^a ) are input, and a coordinate transformation matrix giving the relationship between observed values and catalog values is calculated using the following equation.

Observed values ($ ₁ , $ ₂ , $ ₃ ) and catalog values ($ _1j ^a ,
From $ _2j ^a , $ _3j ^a ) [ _B C _I ^a = [ $ ₁ , $ ₂ , $ ₃ ] [ $ _1j ^a , $ _2j ^a , $ _3j ^a ] ^-1 (9) Observed value ( $ ₁ , $ ₂ , $ ₃ ) and the catalog value ($ _1j ′,
From $ _2j ^a , $ _3j ^a ), [ _B C _I ′] = [$ ₁ , $ ₂ , $ ₃ ] [$ _1j ′, $ _2j ^a , $ _3j ^a ] ^-1 (10) holds.

FIG. 4 shows the relative relationship between the attitude of the artificial satellite at time (n-1)τ and time nτ, assuming that the time t when the star is observed by the star sensor 1 is t=0. However, τ is the sampling time of stellar observation. In the figure, (i _Bo-1 , j _Bo-1 , k _Bo-1 ) is time (n-1) τ
unit vector in the satellite axis direction, (i _Bo , j _Bo ,
k _Bo ) is the unit vector in the satellite axis direction at time nτ, (ΔΨn, Δφn, Δθn) is the Euler angle,
(ω _1o-1 , ω _2o-1 , ω _3o-1 ) and (ω _1o , ω _2o , ω _3o ) are rotational angular velocities around the satellite axis, respectively.

In such a relationship, the posture displacement calculation device 8
calculates the change in posture [〓C _B ] from time t=0 to time t=T using the following formula.

[ΔC _B ] = C ₁₁ C ₁₂ C ₁₃ C 21 C ₂₂ C ₂₃ C ₃₁ C ₃₂ C ₃₃ ……(11) Here, C ₁₁ = Δρ ₁ ² _{−Δρ 2 2} ^−Δρ ₃ ² +Δρ ₄ ² C ₁₂ = 2(-Δρ ₄ Δρ ₃ +Δρ ₁ Δρ ₂ ) C ₁₃ =2(Δρ ₄ Δρ ₂ +Δρ ₁ Δρ ₂ ) C ₂₁ =2(Δρ ₄ Δρ ₃ +Δρ ₁ Δρ ₂ ) C ₂₂ =−Δρ ₁ ² +Δρ ₂ ² +Δρ ₃ ² +Δρ ₄ ² C ₂₃ = 2 (−Δρ ₄ Δρ ₁ +Δρ ₂ Δρ ₃ ) C ₃₁ = 2 (−Δρ ₄ Δρ ₂ +Δρ ₁ Δρ ₃ ) C ₃₂ = 2 (Δρ ₄ Δρ ₁ +Δρ ₂ Δρ ₃ ) _{C 33} = −Δρ ₁ ² −Δρ ₂ ² +Δρ ₃ ² ₊ Δρ ₄ 2However ^, Δρ ₁ Δq _{4 −Δq 3} Δq ₂ Δq ₁ Δρ ₁ Δρ ₂ = Δq ₃ Δq ₄ Δq 1 Δq 2 Δρ 2 Δρ ₃ −Δq ₂ Δq ₁ Δq ₄ Δq ₃ Δρ ₃ Δρ _4o −Δq ₁ −Δq ₂ −Δq ₃ Δq ₄ Δρ _4o-1 (12) Δq ₁ = sinΔθn／2cosΔφn／2sinΔΨn／
2−cosΔθn／2sinΔφn／2cosΔΨn／2 Δq ₂ =−sinΔθn／2cosΔφn／2cosΔΨn
/2-cosΔθn/2sinΔφn/2sinΔΨn/2 Δq ₃ = sinΔθn/2cosΔφn/2cosΔΨn/
2-cosΔθn/2cosΔφn/2sinΔΨn/2 Δq ₄ =cosΔθn/2cosΔφn/2cosΔΨn/
2−sinΔθn／2sinΔφn／2sinΔΨn／2 (13) The unknown quantities here are the Euler angles (Δφ,
Δθ, ΔΨ) _o , but the initial value Δφ ₀ at time t=0
=Δθ ₀ =ΔΨ ₀ =0, rotational angular velocity around the satellite axis measured by inertial sensor 2 ω _uo (u=1, 2, 3)
is read with a period τ and calculated using the following formula.

Δφ _o = 1/6 (d ₁ +2d ₂ +2d ₃ +d ₄ ) Δθ _o = 1/6 (η ₁ +2η ₂ +2η ₃ +η ₄ ) ΔΨ _o = 1/6 (f ₁ +2f ₂ +2f ₃ +f ₄ ) (14 ) However, d ₁ = τω _1o η ₁ = τω _2o f ₁ = τω _3o d ₂ = τ｛ω _1o cos (ω _2o τ/2) + ω _3o sin (ω _2o τ/
2)} η ₂ = τ{ω _1o tan (ω _1o τ/2) sin (ω _2o τ/2)
+ω _2o −ω _3o tan (ω _1o τ/2) cos (ω _2o τ/2)} f ₂ = τ{−ω _1o sin (ω _2o τ/2)/cos (ω _1o τ/2
) +ω _3o cos (ω _2o τ/2)/cos (ω _1o τ/2)} d ₃ = τ{ω _1o cos (η ₂ /2) + ω _3o sin (η ₂ /2)} η ₃ = τ{ ω _1o tan (d ₂ /2) sin (η ₂ /2) + ω _2o −
ω _3o tan (d ₂ /2) cos (η ₂ /2)} f ₃ = τ{−ω _1o sin (η ₂ /2) / cos (d ₂ /2) + ω _3
o cos (η ₂ /2) / cos (d ₂ /2)} d ₄ =τ{ω _1o cosη ₃ +ω _3o sinη} η ₄ =τ{ω _1o tand ₃ sinη ₃ +ω _2o −ω _3o tand ₃ cosη ₃
} f ₄ = τ{−ω _1o sinη ₃ /cosd ₃ +ω _3o cosη ₃ /cosd ₃ } Note that at time t=0 [Δρ ₁ , Δρ ₂ , Δρ ₃ , Δρ ₄ ] ₀ ^T = [0, 0, 0 , 1] ₀ ^T (15).

The posture initial value updating device 9 inputs the output signals [ _B C ^a _I ], [ _B C _I ′] of the matching matrix calculation device 7 and the output signal [〓C _B ] of the posture displacement calculation device 8, and inputs the output signals [ _B C Calculate _I 〕 ₁ = [〓C _B 〓〔 _B C _I ^a 〕 (16) 〔 _B C _I 〕 ₂ = [〓C _B 〓〔 _B C _I ′〕 (17).

With the above, the first processing is completed, and the mode changeover switch 11 is set to M ₁ =M ₂ =1, and the second processing begins.

The star sensor 1 newly observes a star, and the star vector calculator 3 calculates a unit vector in the direction of the star using equation (3).

On the other hand, the sub-catalog editing device 5 edits the sub-catalog in the following manner based on the output signal of the database 4 and the attitude initial value updating device 8.

Referring to equation (4), 〓 _sl1 (t) = [ _S C _B ] _l [ _B C _I ] ₁ [1, 0
, 0] ^T , l = 1, 2 〓 _sl1 (t) = [ _S C _B ] _l [ _B C _I ] ₁ [1, 0
, 0] ^T , l = 1, 2 〓 _sl2 (t) = [ _S C _B ] _l [ _B C _I ] ₂ [1, 0, 0] ^T , l =
Calculate 1, 2(18).

Next, referring to equation (5), from the star catalog ^a in database 4, 〓 _sl1 , 〓 _sl
Edit the subcatalog corresponding to ₂ .

$ ^a _j1 ∈[ ^a | | cos ^-1 { ^a・〓 _sl1 } | <ε ₃ ] $ ^a _j1 ∈〔 ^a | | cos ^-1 { ^a・〓 _sl1 } | <ε ₃ ] $ ^a _j2 ∈〔 ^a ｜｜cos ^-1 { ^a・〓 _sl2 }｜＜ε ₃ 〕(1
9) The star identification processing device 6 generates the unit vectors $ ₁ , $ ₂ , $ ₃ in the direction of the observed star and the subcatalog $ _j1 ^a ,
Perform the following identification process using $ _j2 ^a , [ _B C _I ] ₁ , and [ _B C _I ] ₂ as input. First, [$ ₁₁ , $ ₁₂ , $ ₁₃ ] = [ _I C _B ] ^T ₁ [$ ₁ , $ ₂ , $ ₃ ], (20) T: Transposition [$ ₂₁ , $ ₂₂ , $ ₂₃ ] = [ _I C _B ] ^T ₂ [$ ₁ , $ ₂ , $ ₃ ], (21) T: Calculate transpose.

Next, for the small value ε ₄ , for all stars in each subcatalog $ ₁₁・$ _j1 ^a ＞cosε ₄ $ ₁₂ , $ _j1 ^a ＞cosε ₄ $ ₁₃・$ _j1 ^a ＞cosε ₄ (22 ) $ ₂₁・$ _j2 ^a ＞cosε ₄ $ ₂ B・$ _j2 ^a ＞cosε ₄ $ ₂₃・$ _j3 ^a ＞cosε ₄ (23) Test to see if there is a combination that satisfies the following. Remove from subcatalog. As a result, for example, the three stars $ ^a _1j1 , $ ^a _2j1 , and $ ^a _3j1 ⁱⁿ the subcatalog $ _j1 a each cost $ ₁₁ ,
It remains in the catalog as a pair corresponding to $ ₁₂ and $ ₁₃ . At this time, subcatalog S _j2 ^a is generally completely removed.

Based on this process, the identification processing device 6 outputs [ _B C _I ] ₁ .

The posture initial value updating device 9 is the posture displacement calculation device 8.
The change in posture from time t=T to t=2T calculated by equations (11) to (15) in [C _B ] and the output [ _B C _I ] ₁ of the identification processing device 6 are input, and the time t is calculated.
= Relationship between the satellite attitude at 2T, that is, the satellite axis coordinates (X _B , Y _B , Z _B ) and the inertial space coordinates (X _I , Y _I , Z _I ) as the attitude star reference coordinates [ _B C _I ] _2T Calculate and output using the following formula.

[ _B C _I ] _2T = [〓C _B ] [ _B C _I ] ₁ (24) The attitude calculation device 10 uses the calculation result of the above equation (24) until the attitude initial value is updated again after 2T time. As an initial value, the posture is calculated and output using the following formula.

Here, if the update time T=2T of the initial attitude value is set to t=0 again, then the attitude of the satellite [〓 _B , 〓 _B ,
〓 _B 〕 _k is [i _B , j _B , k _B ] ^T = [ _B C _I ] _k [i _I , j _I , k _I ] ^T (25) [ _B C _I ] _k = [〓C _B ] [ _B C _I ] _k-1 (26) However, for [ _B C _I ] ₀ , the calculation result of equation (24) is used. In addition, [〓C _B ] reads inertial sensor data ω _l (l = 1, 2, 3) every time τ = t _k - t _k-1 , and uses the same formula as equations (14) and (15) to calculate the oil value. Calculate the angles Δφ, Δθ, ΔΨ,
Substitute this into equation (13), calculate the Euler parameters (Δρ ₁ , Δρ ₂ , Δρ ₃ , Δρ ₄ ) from equations (12) and (13), and substitute this into the right side of equation (11). Calculate [〓C _B ] sequentially.

As is clear from the above, the attitude determination method according to the present invention is based on the attitude at the time of star observation obtained based on star identification, and the change in attitude during the star identification processing period obtained using inertial sensor data. The initial value of the attitude is updated regularly by combining the two, and the attitude of the satellite can be determined sequentially based on this. Therefore, a filter for posture estimation is not required. In addition, the output frequency of attitude determination is high. Also,
Using equations (6) and (7), the observed star can be uniquely identified regardless of the magnitude of the error in the attitude information. Furthermore, since the initial posture value is updated based on the results of identification processing performed multiple times (twice in this embodiment), malfunctions of the system due to identification errors can be prevented. It has the following advantages.

Note that this example shows a method of directly using stars in the subcatalog to process equation (7), but in order to reduce the number of vector calculations (inner products) using equation (7), the observed values and stars in the subcatalog ^Among _{the stars of _ _} ^_ _{_ _ _} ^_ _{_ _} Even if the core catalog is edited and formula (7) is applied to the stars in this core catalog, the effects of the present invention will not be hindered in any way.

[Brief explanation of drawings]

Fig. 1 is a diagram showing the configuration concept of the attitude determination system according to the present invention, Fig. 2 is a conceptual diagram showing the installation of a stellar sensor, Fig. 3 is a diagram showing a mathematical model for star observation, and Fig. 4 is a diagram showing the sequential attitude determination system. It is a diagram explaining the relationship between the previous stage attitude and the current attitude in the process of change calculation, 1 is a stellar sensor, 2 is an inertial sensor, 3 is a stellar vector calculation device, 4 is a database, 5 is a sub-catalog editing device, 6 is a star identification processing device, 7 is a matching matrix calculation device, 8 is an attitude displacement calculation device, 9 is an attitude initial value update device, 10 is an attitude calculation device, 11 is a mode changeover switch, and 12 is an attitude determination computer. .

Claims (1)

[Claims] 1 A stellar vector calculation device that calculates a unit vector in the stellar direction with respect to the satellite axis coordinates using data from a stellar sensor mounted on the satellite as input, and an attitude prediction device used at the stage before obtaining a stellar catalog and attitude determination results. A sub-catalog editing device that reads values from the database and edits a catalog of stars predicted to be within the field of view of the star sensor, and a sub-catalog editing device that uses the output signals of the above-mentioned stellar vector calculation device and sub-catalog editing device as input to correspond to observed stars. An identification processing device that identifies catalog stars, a matching matrix calculation device that uses the output signal of the identification processing device as an input signal to calculate a correspondence matrix between observed values and their catalog values, and data from an inertial sensor mounted on an artificial satellite. an attitude displacement calculation device that calculates the amount of change in attitude after a certain period of time from the time of star observation by the stellar sensor with input as input; and an attitude displacement calculation device that uses as input the output signals of the above-mentioned matching matrix calculation device and attitude displacement calculation device to calculate the initial value of the attitude. an attitude initial value update device; an attitude calculation device that calculates the attitude of an artificial satellite by inputting the above attitude initial value update device and inertial sensor data; A mode switch is provided for the first processing, and in the first processing, the result of the star identification processing is input into the above matching matrix, and the update result of the attitude initial value is input into the subcatalog editing device, and in the second processing, the result of the stellar identification processing is input into the matching matrix. An attitude determining computer configured to input the update result of the initial attitude value to the attitude initial value updating device, input the update result of the initial attitude value to the attitude calculating device, and feed back the output signal of the attitude calculating device to the subcatalog editing device. An attitude determination method for an artificial satellite, characterized in that the method is mounted on the above artificial satellite and outputs an attitude determination value and an attitude change rate of the artificial satellite.