CN102999042A - Layering fault autonomic diagnostic method of global navigation chart (GNC) system of deep space probe - Google Patents

Abstract

Provided is a layering fault autonomic diagnostic method of a global navigation chart (GNC) system of a deep space probe. The layering fault autonomic diagnostic method comprises the steps of (1) establishing a fault-measurement point incidence matrix, determining that a fault mode is the diagnostic result if the fault mode can be determined only according to the fault-measurement point incidence matrix, and executing the step (2) to perform component-level fault diagnosis if the fault mode can not be determined only according to the fault-measurement point incidence matrix; (2) performing fault diagnosis by utilizing the redundancy relation between sensors of the GNC system and consistency of the input and output relation of an executing mechanism to obtain the member having the fault, and executing the step (3) to perform system-level fault diagnosis if the redundancy relation is not met; (3) judging whether the GNC system is the smallest system, calculating the theoretical angular velocity of the probe if the GNC system is not the smallest system, diagnosing out the specific member having the fault according to the consistency of the theoretical angular velocity and the angular velocity measured by the sensors, and executing the step (4) if the GNC system is the smallest system; and (4) accumulating air injection time of a thruster in an X-axis direction, a Y-axis direction and a Z-axis direction, determining that the GNC system has the fault if the accumulated time in any of the X-axis direction, the Y-axis direction and the Z-axis direction exceeds the preset threshold value within the fixed time, and determining that the GNC system is normal if the accumulated time in any of the X-axis direction, the Y-axis direction and the Z-axis direction does not exceed the preset threshold value within the fixed time.

Description

Deep space probe GNC system layer method for automatic fault diagnosis

Technical field

The invention belongs to the Spacecraft malfunction process field, relate to a kind of method for automatic fault diagnosis, be applicable to the autonomous diagnosis of fault and processing of deep space probe GNC system.

Background technology

For the survey of deep space task, the detected object of spacecraft, purpose and residing environment all are different from earth satellite system, thereby operation and the control technology of spacecraft have been brought new challenge.At first the deep space probe flight time long, the deep space environment X factor is many and complicated, this is just so that the probability increase that GNC system and parts meet with accident and break down.Secondly, the communication delay of deep space probe and ground control station is large, and signal also may be blocked by the sun and other celestial bodies, and this is so that slow with the control reaction based on the navigation of ground control station, being unfavorable for the processing of accident, will be very dangerous for manned survey of deep space task especially.Therefore, in order to guarantee can in time to process after deep space probe breaks down, the Reduction of failure risk needs development automatic fault diagnosis technology, be implemented in the situation that ground communication interrupts fully and still can independently to detection, isolation and the location of fault, strengthen the autonomous viability of survey of deep space.For this reason, NASA (NASA) and European Space Agency (ESA) have all considered employing automatic fault diagnosis technology in detection mission, corresponding automatic fault diagnosis system has all been developed in Rosetta (Rosetta) plan of surveying such as Saturn detector Cassini, NEAR task (NEAR), Deep Space 1 (DS-1), the plan of deep space bump and comet etc.China has also considered the automatic fault diagnosis technology at the deep space probe design aspect, but the trouble diagnosibility that has at present a little less than, not yet form the method for diagnosing faults of system, and independence is inadequate.

Summary of the invention

Technology of the present invention is dealt with problems and is: overcome the deficiencies in the prior art, a kind of deep space probe GNC system layer method for automatic fault diagnosis is provided, the method is by being divided into fault diagnosis several levels such as component-level fault diagnosis, component level fault diagnosis, Methods for Diagnosing System Level Malfunctions, thereby so that fault can independently be diagnosed after occuring as early as possible, Security of the system and reliability have been guaranteed.

Technical solution of the present invention is: deep space probe GNC system layer method for automatic fault diagnosis, and step is as follows:

(1) according to unit failure pattern analysis (FMEA) result, sets up fault-measuring point incidence matrix; The self check information that provides according to the GNC various parts, analog quantity telemetry intelligence (TELINT) and carry out fault diagnosis according to fault-measuring point incidence matrix, if can unique definite fault mode according to fault-measuring point incidence matrix, then this fault mode be diagnostic result; Otherwise turn step (2) and carry out the component level fault diagnosis;

(2) utilize redundancy relationship between the GNC system sensor and the consistance of topworks's input/output relation to carry out fault diagnosis, the parts that obtain breaking down; Turn step (3) and carry out Methods for Diagnosing System Level Malfunctions not satisfying redundancy relationship or parts that can not unique definite fault;

(3) judge whether the GNC system is minimum system, if not minimum system, then at first utilize the redundancy relationship of sensor or topworks to judge it is sensor failure or actuator failure, dynamics and the kinematical equation of the theoretical control moment substitution detector that again controller is produced, resolve the theoretical angular velocity of detector, diagnose out the parts that specifically break down according to the consistance of theoretical angular velocity and sensor measured angular speed; Otherwise turn step (5); Described minimum system is that sensor, topworks do not exist redundancy;

(4) jet time of accumulative total thruster three axle all directions, in setting time, if the accumulative total jet time of three axle either directions surpasses predefined threshold value, then there is fault in the GNC system, otherwise the GNC system is normal.

Described component level fault diagnosis mainly comprises based on gyro diagnosis, star sensor diagnosis, gyro and the star sensor in odd even space unites diagnosis; Momenttum wheel fault diagnosis based on the input and output direct redundancy.

Described gyro and star sensor are united diagnosis and are applicable to gyro and star sensor quantity summation more than or equal to 5 o'clock, and hypothesis generation single fault, and diagnosis algorithm is as follows:

(2.1) when gyro quantity is 4, according to the consistance of each gyro output, judge whether gyro is unusual, if output is consistent, then gyro is all normal, otherwise the gyro existence is unusual, turns step (2.3);

(2.2) when star sensor quantity is 2, according to the consistance of each star sensor output, judge whether star sensor is unusual, if output is consistent, then star sensor is normal, otherwise star sensor is unusual, turns step (2.3);

(2.3) exist when unusual when gyro or any one parts of star sensor, the consistance of the angular velocity that then records according to gyro and star sensor is determined trouble unit.

The present invention compared with prior art beneficial effect is:

(1) the survey of deep space spacecraft method for automatic fault diagnosis of the present invention's proposition has adopted layering, at first realizes the larger fault of parts is in time detected isolation by the component-level diagnosis, has guaranteed parts and security of system; Then carry out component level diagnosis, can realize that small fault to parts detects and isolates; When the parsing redundancy of system is not enough to carry out the component level diagnosis, introduce dynamics and kinematics, take full advantage of system-level redundancy relationship and carry out system level diagnostic, can realize the fault isolation to sensor and topworks; When the GNC system is minimum system when (there are not redundancy in sensor and topworks), whether surpass the setting threshold detection system by thruster accumulative total jet time at the appointed time and whether have fault.Above-mentioned level and is limited in as far as possible little scope with fault effects so that fault can be diagnosed after occuring as early as possible, and processing for consequent malfunction provides condition, can significantly improve fault detect rate and the isolation rate of deep space probe.

(2) the component-level fault diagnosis processor that can take full advantage of parts self carries out the data validity interpretation, reduces the calculating pressure of spaceborne computer;

(3) the component level fault diagnosis takes full advantage of redundancy relationship and the input/output relation between the sensor, and calculated amount is little, real-time good.

(4) Methods for Diagnosing System Level Malfunctions utilizes dynamics and kinematics to refine system-level parsing redundancy relationship, can realize the fault isolation of sensor and topworks.

(5) layering that adopts of the present invention and is limited in as far as possible little scope with fault effects so that fault can be diagnosed after occuring as early as possible, has guaranteed Security of the system and reliability, meets the engineering actual demand.

Description of drawings

Fig. 1 is the inventive method process flow diagram.

Embodiment

The invention provides a kind of deep space probe GNC system method for automatic fault diagnosis, the fault diagnosis of GNC system has been divided into component-level fault diagnosis, component level fault diagnosis, three levels of Methods for Diagnosing System Level Malfunctions.At first utilize self check information and the analog quantity telemetry of parts self to carry out the component-level fault diagnosis, next utilizes redundancy relationship, input/output relation between the parts to carry out the component level fault diagnosis, along with failure condition worsens, when the component-level redundancy relationship can not satisfy the redundant demand of component level diagnosis, utilize spacecraft dynamics to carry out Methods for Diagnosing System Level Malfunctions, when the GNC system is minimum system, whether surpasses the setting threshold detection system by thruster accumulative total jet time at the appointed time and whether have fault at last.Below in conjunction with accompanying drawing the specific embodiment of the present invention is further described in detail.

(1) component-level fault diagnosis

Self check information and analog quantity telemetry that this step utilizes parts self to provide adopt the fault dictionary method to diagnose, and can navigate to the functional module of parts.The self check information of parts comprises RAM self check, reset mode, data effective marker, sees high light sign, mode of operation etc., and is generally normal or unusual with 0 or 1 expression.The analog quantity telemetry comprises power supply remote measurement, remote temperature sensing, motor remote measurement, telemetering of current, rotary speed direction remote measurement etc., is generally the voltage of 0～5V.

For the analog quantity remote measurement, at first be converted into the form of 0 or 1 expression by the red line method.For example, the power supply remote measurement is that 4～5V is effective, then when measured value V ∈ [4,5], thinks that the power supply remote measurement is normal, represents with 0; When measured value V ∈ [0,4) time, think that the power supply remote measurement is unusual, represent with 1.

According to unit failure pattern analysis (FMEA) result or fault simulation analysis result, set up fault-measuring point incidence matrix, as shown in table 1.

Table 1 fault-measuring point incidence matrix

	Measuring point 1	Measuring point 2	…	Measuring point n
					Fault mode 1	1	0	…	1
Fault mode 2	0	0	…	1
					…	…	…	…	…
Fault mode m	0	1	…	0

In the upper table, behavior measuring point (set of parts self check information and analog quantity telemetry intelligence (TELINT)), classify fault mode as, the numeral fault mode of ranks infall and the incidence relation of measuring point, be 0 expression fault mode on measuring point without impact, be that measuring point showed as unusually after 1 expression fault mode occured.Therefore, formed fault dictionary according to fault-measuring point incidence matrix, in the output of detector real-time each measuring point of detection in service, when the measuring point output abnormality, according to the abnormal conditions of different measuring points, by the fault dictionary shown in 1 of tabling look-up, just can realize the component-level fault diagnosis.For example, when detecting measuring point 2 unusual (numeral of the row of corresponding measuring point 2 is 1), all when normal (numeral of corresponding other measuring points is 0), can obtain fault by the fault dictionary shown in the table 1 is fault mode n to other measuring points.

(2) component level fault diagnosis

When component-level diagnosis can not unique definite source of trouble, then be transferred to the component level diagnosis.The component level fault diagnosis mainly comprise similar sensor mutual diagnosis, inhomogeneity sensor unite diagnosis, based on the conforming actuator failure diagnosis of input and output.For deep space probe GNC system configuration commonly used at present, here influences based on gyro diagnosis, star sensor diagnosis, gyro and the star sensor in odd even space unite diagnosis, based on the conforming momenttum wheel fault diagnosis of input and output, and actual according to engineering, think simultaneously a fault only to occur, namely satisfy the single fault hypothesis.

The component level diagnosis mainly realizes by the consistance of the redundancy relationship between the sensor and topworks's input/output relation, therefore needs to satisfy following conditions for diagnostics:

When 1. utilizing a plurality of gyros to diagnose mutually, require to participate in and decide the gyro number of appearance greater than 4;

When 2. utilizing a plurality of star sensors to diagnose mutually, require to participate in and decide the star sensor number of appearance greater than 2;

3. utilize gyro and star sensor to unite when diagnosis, require to participate in the gyro of deciding appearance and star sensor quantity summation more than or equal to 5;

4. controlling computing machine can obtain the steering order and the momenttum wheel rotating speed that send to momenttum wheel, the duty such as turn to.

Satisfying on the basis of above condition, realizing the component level diagnosis by following steps:

1. when participating in the gyro number during greater than 4 decide appearance, utilize a plurality of gyros to diagnose mutually

When available gyro number during greater than 4, introduce wherein near 90 ° principle as far as possible according to angle between the gyro to measure axle that 5 gyros are the work gyro, the measured angular speed of establishing respectively 5 gyros is g ₁, g ₂..., g ₅, matrix is installed is respectively A ₁, A ₂..., A ₅, A ₁～A ₅Be the vector of 1 * 3 dimension.Then by any 3 gyro i wherein, j, (i ≠ j ≠ k) definite detector three axis angular rates are k

${\mathrm{\ω}}_{\mathrm{ijk}}=\mathrm{inv}\left(\left[\begin{array}{c}{A}_i\\ A_j\\ A_k\end{array}\right]\right)\left[\begin{array}{c}{g}_i\\ g_j\\ g_k\end{array}\right]---\left(1\right)$

Wherein, inv () represents matrix inversion (together lower).Then measured value and the ω of the individual gyro of l (l ≠ i, j, k) _IjkResidual error between the axial projection of this gyro to measure is

ε _ijkl＝|g _l-A _lω _ijk| (2)

With i, j, k, l then can obtain respectively ε respectively in 1～5 value ₁₂₃₄, ε ₁₂₃₅, ε ₂₃₄₅, ε ₁₃₄₅, ε ₁₂₄₅, set threshold residual value r ₀, so that when i gyro failure, residual error relevant with i in the footnote is all greater than r ₀, and with the irrelevant residual error of i less than r ₀

Detector in orbit in, utilize residual epsilon _IjklWith threshold value r ₀Realization is to the diagnosis of gyro: respectively gyro 1～5 is set diagnosis score value f ₁～f ₅, work as ε _Ijkl＞r ₀The time, f then _i, f _j, f _k, f _lAll subtract 1, work as ε _Ijkl＜r ₀The time, f then _i, f _j, f _k, f _lAll add 1, it is the fault gyro that the diagnosis score value reaches 0 gyro at first.

2. when participating in the star sensor number during greater than 2 decide appearance, utilize a plurality of star sensors to diagnose mutually

When the quick number of available star during greater than 2, optional wherein three stars are quick quick for the work star, establish respectively 3 quick optical axises that record of star at the Z that is oriented to of inertial system _S1, Z _S2, Z _S3, then calculate respectively three optical axis included angles that star is quick according to measured value:

${\mathrm{\α}}_{\mathrm{ij}}=\arccos \left(\frac{Z_{\mathrm{Si}}\·Z_{\mathrm{Sj}}}{\left|Z_{\mathrm{Si}}\right|\left|Z_{\mathrm{Sj}}\right|}\right)$ I ≠ j and i, j ∈ (1,2,3) (3)

The theoretical angle that can obtain respectively between the optical axis according to the quick installation matrix of star is α _Ij0, then measurement result and the residual error between the theoretical value of angle are between the quick i of star and the quick j of star:

ε _ij＝|α _ij-α _ij0| (4)

Set threshold residual value r _S0, so that when i the quick fault of star, residual error relevant with i in the footnote is all greater than r _S0, and with the irrelevant residual error of i less than r _S0

Detector in orbit in, utilize residual epsilon _IjWith threshold value r _S0The diagnosis that realization is quick to star: respectively star quick 1～3 is set diagnosis score value f _S1～f _S3, work as ε _Ij＞r _S0The time, f then _Si, f _SjAll subtract 1, work as ε _Ij＜r _S0The time, f then _Si, f _SjAll add 1, the diagnosis score value reach at first 0 star quick for the fault star quick.

3. when not satisfying 1. and requirement 2. and participate in the gyro of deciding appearance and star sensor summation during more than or equal to 5, utilize gyro and star sensor to unite diagnosis

Need to unite comprising of diagnosis of following 2 kinds of situations:

A.4 gyro+1 star sensor

When the gyro number be 4, when the quick number of star is 1, at first whether fault detects to gyro.Calculate ε according to formula (2) ₁₂₃₄If, ε ₁₂₃₄＜r ₀Illustrate that then 4 gyros are all working properly, if otherwise ε ₁₂₃₄＞r ₀Then there is gyro failure in explanation.

If gyro is all normal, then can be by the gyro to measure result to the quick fault isolation of carrying out of star.If the quick three-axis attitude angular velocity that records of star is ω _S=[ω _Sx, ω _Sy, ω _Sz] ', utilizes gyro to calculate three-axis attitude angular velocity

${\mathrm{\ω}}_{1234}=\left[\begin{array}{c}{\mathrm{\ω}}_{1234x}\\ {\mathrm{\ω}}_{1234y}\\ {\mathrm{\ω}}_{1234z}\end{array}\right]=\mathrm{inv}\left(\left[\begin{array}{cccc}{A}_1^{\′}& A_2^{\′}& A_3^{\′}& A_4^{\′}\end{array}\right]\left[\begin{array}{c}{A}_1\\ A_2\\ A_3\\ A_4\end{array}\right]\right)\left[\begin{array}{cccc}{A}_1^{\′}& A_2^{\′}& A_3^{\′}& A_4^{\′}\end{array}\right]\left[\begin{array}{c}{g}_1\\ g_2\\ g_3\\ g_4\end{array}\right]---\left(5\right)$

Then calculate residual error:

${\mathrm{\ϵ}}_S=\left|\underset{j=1}{\overset{m}{\mathrm{\Σ}}}({\mathrm{\ω}}_{\mathrm{Sx}}(t_0+j*\mathrm{\Δt})-{\mathrm{\ω}}_{1234x}(t_0+j*\mathrm{\Δt}))\right|+\left|\underset{j=1}{\overset{m}{\mathrm{\Σ}}}({\mathrm{\ω}}_{\mathrm{Sy}}(t_0+j*\mathrm{\Δt})-{\mathrm{\ω}}_{1234y}(t_0+j*\mathrm{\Δt}))\right|$

$+\left|\underset{j=1}{\overset{m}{\mathrm{\Σ}}}({\mathrm{\ω}}_{\mathrm{Sz}}(t_0+j*\mathrm{\Δt})-{\mathrm{\ω}}_{1234z}(t_0+j*\mathrm{\Δt}))\right|---\left(6\right)$

Δ t control cycle in the formula, t ₀For calculating initial time, m is cumulative number, and subscript " ' " represents matrix transpose (together lower).

Work as ε _SGreater than setting threshold r _S0The time, the quick fault of star then.

If there is fault in gyro, under the single fault hypothesis, by the star sensor measurement result it is isolated.Calculate respectively residual error

${\mathrm{\ϵ}}_{\mathrm{Gi}}=|\underset{j=1}{\overset{m}{\mathrm{\Σ}}}{g}_i-A_i\underset{j=1}{\overset{m}{\mathrm{\Σ}}}({\mathrm{\ω}}_S(t_0+j*\mathrm{\Δt})*\mathrm{\Δt})|$ i＝1，2，3，4 (7)

Work as ε _GiGreater than setting threshold r _G0The time, gyro i fault then.

B.3 gyro+2 star sensors

When the gyro number be 3, when the star sensor number is 2, at first whether fault detects to star sensor.If the quick three-axis attitude angular velocity that records of star is respectively ω _S1=[ω _S1x, ω _S1y, ω _S1z] ' and ω _S2=[ω _S2x, ω _S2y, ω _S2z] ', calculates ε according to formula (4) ₁₂If, ε ₁₂＜r _S0Illustrate that then two stars are quick working properly, if otherwise ε ₁₂＞r _S0Then there is the quick fault of star in explanation.

If star sensor is all normal, can carry out fault detect and isolation to gyro by the star sensor measurement result, calculate respectively residual error

${\mathrm{\ϵ}}_{\mathrm{Gi}}=|\underset{j=1}{\overset{m}{\mathrm{\Σ}}}{g}_i-A_i\underset{j=1}{\overset{m}{\mathrm{\Σ}}}({\mathrm{\ω}}_{S1}(t_0+j*\mathrm{\Δt})+{\mathrm{\ω}}_{S2}(t_0+j*\mathrm{\Δt}))*\mathrm{\Δt}/2|$ i＝1，2，3 (8)

Work as ε _GiGreater than setting threshold r _G0The time, gyro i fault then.

If there is fault in star sensor, under the single fault hypothesis, the result isolates it by gyro to measure.Utilize the measured value calculating detector three axis angular rate ω of 3 gyros according to formula (1) _Ijk=[ω _Ijkxω _Ijkyω _Ijkz] ', then calculates respectively residual error:

${\mathrm{\ϵ}}_{\mathrm{Si}}=\left|\underset{j=1}{\overset{m}{\mathrm{\Σ}}}({\mathrm{\ω}}_{\mathrm{Six}}(t_0+j*\mathrm{\Δt})-{\mathrm{\ω}}_{\mathrm{ijkx}}(t_0+j*\mathrm{\Δt}))\right|+\left|\underset{j=1}{\overset{m}{\mathrm{\Σ}}}({\mathrm{\ω}}_{\mathrm{Siy}}(t_0+j*\mathrm{\Δt})-{\mathrm{\ω}}_{\mathrm{ijky}}(t_0+j*\mathrm{\Δt}))\right|$

$+\left|\underset{j=1}{\overset{m}{\mathrm{\Σ}}}({\mathrm{\ω}}_{\mathrm{Siz}}(t_0+j*\mathrm{\Δt})-{\mathrm{\ω}}_{\mathrm{ijkz}}(t_0+j*\mathrm{\Δt}))\right|$ i＝1，2 (9)

Δ t control cycle in the formula, t ₀For calculating initial time, m is cumulative number.

Work as ε _SiGreater than setting threshold r _S0The time, the quick i fault of star then.

4. diagnose based on the momenttum wheel of input/output relation

If t constantly momenttum wheel input control order is U _w(t), then the theoretical variable quantity of interior momenttum wheel angular momentum of m* Δ t time is $\mathrm{\ΔH}=\underset{i=1}{\overset{m}{\mathrm{\Σ}}}(\mathrm{\ΔU}(t_0+i*\mathrm{\Δt})*K),$ Wherein Δ U (t) inputs the variable quantity of instruction constantly to t for t-1.The theoretical variable quantity that therefore can calculate the momenttum wheel rotating speed is Δ ω _w=Δ H/I _wObtain residual error according to momenttum wheel actual change amount and theoretical variable quantity

ε _w＝|Δω _w-(ω _w(t ₀+mΔt)-ω _w(t ₀))| (10)

Work as ε _wGreater than threshold value r _W0The time, then think the momenttum wheel fault.

(3) Methods for Diagnosing System Level Malfunctions

When not satisfying the redundant condition of component level diagnosis, need to introduce star dynamics and the kinematics of detector, utilize the parsing redundancy between the mathematical model of sensor, topworks, controller and dynamics, the kinematics to carry out Methods for Diagnosing System Level Malfunctions.Because momenttum wheel can be diagnosed by the direct redundancy based on input and output, does not often need to be placed on system-level the diagnosis.

System level diagnostic is often used for not satisfying the component level conditions for diagnostics, and the GNC system is not the situation of minimum system.Here minimum system refers to the system that there are not redundancy in sensor, topworks.As a kind of typical case, analyzed here that sensor comprises 3 gyros, 1 star is quick and the situation of 1 group of thruster (containing the positive negative direction of three axles), other situations can utilize above-mentioned thought to carry out similarity analysis.

For 3 gyro+1 stars quick+situation of 1 group of thruster (containing the positive negative direction of three axles), at first utilize the measured value calculating detector three axis angular rate ω of 3 gyros according to formula (1) _Ijk=[ω _Ijkxω _Ijkyω _Ijkz] ', detects sensor by following formula again and whether has fault:

${\mathrm{\ϵ}}_S=\left|\underset{j=1}{\overset{m}{\mathrm{\Σ}}}({\mathrm{\ω}}_{\mathrm{Sx}}(t_0+j*\mathrm{\Δt})-{\mathrm{\ω}}_{\mathrm{ijkx}}(t_0+j*\mathrm{\Δt}))\right|+\left|\underset{j=1}{\overset{m}{\mathrm{\Σ}}}({\mathrm{\ω}}_{\mathrm{Sy}}(t_0+j*\mathrm{\Δt})-{\mathrm{\ω}}_{\mathrm{ijky}}(t_0+j*\mathrm{\Δt}))\right|$

$+\left|\underset{j=1}{\overset{m}{\mathrm{\Σ}}}({\mathrm{\ω}}_{\mathrm{Sz}}(t_0+j*\mathrm{\Δt})-{\mathrm{\ω}}_{\mathrm{ijkz}}(t_0+j*\mathrm{\Δt}))\right|---\left(11\right)$

Δ t control cycle in the formula, t ₀For calculating initial time, m is cumulative number.

Work as ε _SGreater than setting threshold r _S0The time, then there is fault in sensor, otherwise sensor is normal.

On this basis, drawing-in system kinetics equation:

$I_x{\mathrm{\ω}}_x=\frac{1}{S}[(I_y-I_z){\mathrm{\ω}}_y{\mathrm{\ω}}_z+u_{x,r}]$

$I_y{\mathrm{\ω}}_y=\frac{1}{S}[(I_z-I_x){\mathrm{\ω}}_z{\mathrm{\ω}}_x+u_{y,r}]---\left(12\right)$

$I_z{\mathrm{\ω}}_z=\frac{1}{S}[(I_x-I_y){\mathrm{\ω}}_x{\mathrm{\ω}}_y+u_{z,r}]$

u _x，r＝u _x+Δu _x

u _y，r＝u _y+Δu _y (13)

u _z，r＝u _z+Δu _z

I=diag (I wherein _x, I _y, I _z) be detector inertia battle array, u _r=[u _{X, r}u _{Y, r}u _{Z, r}] ' be working control moment, u=[u _xu _yu _z] ' for the desired control moment that controller sends, can be obtained by controller.Δ u=[Δ u _xΔ u _yΔ u _z] ' be the deviation between working control moment and the desired control moment, normal condition is in a small amount.

When there is fault in sensor, make Δ u=0, with u=[u _xu _yu _z] ' substitution formula (12) calculates the theoretical angular velocity omega of three axles _t=[ω _Tx, ω _Ty, ω _Tz] ', is by following formula isolated fault sensor:

${\mathrm{\ϵ}}_{\mathrm{Gi}}=|\underset{j=1}{\overset{m}{\mathrm{\Σ}}}{g}_i-A_i\underset{j=1}{\overset{m}{\mathrm{\Σ}}}{\mathrm{\ω}}_t(t_0+j*\mathrm{\Δt})*\mathrm{\Δt}|$ i＝1，2，3 (14)

${\mathrm{\ϵ}}_S=\left|\underset{j=1}{\overset{m}{\mathrm{\Σ}}}({\mathrm{\ω}}_{\mathrm{Sx}}(t_0+j*\mathrm{\Δt})-{\mathrm{\ω}}_{\mathrm{tx}}(t_0+j*\mathrm{\Δt}))\right|+\left|\underset{j=1}{\overset{m}{\mathrm{\Σ}}}({\mathrm{\ω}}_{\mathrm{Sy}}(t_0+j*\mathrm{\Δt})-{\mathrm{\ω}}_{\mathrm{ty}}(t_0+j*\mathrm{\Δt}))\right|$

$+\left|\underset{j=1}{\overset{m}{\mathrm{\Σ}}}({\mathrm{\ω}}_{\mathrm{Sz}}(t_0+j*\mathrm{\Δt})-{\mathrm{\ω}}_{\mathrm{tz}}(t_0+j*\mathrm{\Δt}))\right|---\left(15\right)$

Δ t control cycle in the formula, t ₀For calculating initial time, m is cumulative number.

Work as ε _GiGreater than setting threshold r _G0The time, then there is fault in i gyro.

Work as ε _SGreater than setting threshold r _S0The time, then there is fault in star sensor.

When sensor is all normal, make Δ u=0, with u=[u _xu _yu _z] ' substitution formula (12) calculates the theoretical angular velocity omega of three axles _t=[ω _Tx, ω _Ty, ω _Tz] ', is by following formula isolated fault sensor:

${\mathrm{\ϵ}}_T=\left[\begin{array}{c}{\mathrm{\ϵ}}_{\mathrm{Tx}}\\ {\mathrm{\ϵ}}_{\mathrm{Ty}}\\ {\mathrm{\ϵ}}_{\mathrm{Tz}}\end{array}\right]=|\underset{j=1}{\overset{m}{\mathrm{\Σ}}}{\mathrm{\ω}}_t(t_0+j*\mathrm{\Δt})-\underset{j=1}{\overset{m}{\mathrm{\Σ}}}\mathrm{inv}\left(\left[\begin{array}{c}{A}_i\\ A_j\\ A_k\end{array}\right]\right)\left[\begin{array}{c}{g}_i(t_0+j*\mathrm{\Δt})/\mathrm{\Δt}\\ g_j(t_0+j*\mathrm{\Δt})/\mathrm{\Δt}\\ g_k(t_0+j*\mathrm{\Δt})/\mathrm{\Δt}\end{array}\right]|---\left(16\right)$

Work as ε _Tx, ε _Ty, ε _TzGreater than setting threshold r _T0The time, then there is fault in the thruster on the respective direction.

(4) fault detect that retrains based on jet time

When surveying as minimum system, only carry out fault detect, no longer carry out fault isolation.During detector stable operation, with t ₀Constantly begin the jet time of the positive negative direction of three axles in the accumulative total m* Δ t time, if the accumulative total jet time of any direction surpasses setting threshold r _T0, then there is fault in system.

The content that is not described in detail in the instructions of the present invention belongs to those skilled in the art's known technology.

Claims (3)

1. deep space probe GNC system layer method for automatic fault diagnosis is characterized in that step is as follows:

(1) according to unit failure pattern analysis (FMEA) result, sets up fault-measuring point incidence matrix; The self check information that provides according to the GNC various parts, analog quantity telemetry intelligence (TELINT) and carry out fault diagnosis according to fault-measuring point incidence matrix, if can unique definite fault mode according to fault-measuring point incidence matrix, then this fault mode be diagnostic result; Otherwise turn step (2) and carry out the component level fault diagnosis;

(2) utilize redundancy relationship between the GNC system sensor and the consistance of topworks's input/output relation to carry out fault diagnosis, the parts that obtain breaking down; If not satisfying redundancy relationship turns step (3) and carries out Methods for Diagnosing System Level Malfunctions;

(3) judge whether the GNC system is minimum system, if not minimum system, then at first utilize the redundancy relationship of sensor or topworks to judge it is sensor failure or actuator failure, dynamics and the kinematical equation of the theoretical control moment substitution detector that again controller is produced, resolve the theoretical angular velocity of detector, diagnose out the parts that specifically break down according to the consistance of theoretical angular velocity and sensor measured angular speed; Otherwise turn step (4); Described minimum system is that sensor, topworks do not exist redundancy;

(4) jet time of accumulative total thruster three axle all directions, in setting time, if the accumulative total jet time of three axle either directions surpasses predefined threshold value, then there is fault in the GNC system, otherwise the GNC system is normal.

2. deep space probe GNC system layer method for automatic fault diagnosis according to claim 1 is characterized in that: described component level fault diagnosis mainly comprises based on gyro diagnosis, star sensor diagnosis, gyro and the star sensor in odd even space unites diagnosis and based on the momenttum wheel fault diagnosis of input and output direct redundancy.

3. deep space probe GNC system layer method for automatic fault diagnosis according to claim 2, it is characterized in that: described gyro and star sensor are united diagnosis and are applicable to gyro and star sensor quantity summation more than or equal to 5 o'clock, and suppose single fault occurs, diagnosis algorithm is as follows:

(2.1) when gyro quantity is 4, according to the consistance of each gyro output, judge whether gyro is unusual, if output is consistent, then gyro is all normal, otherwise the gyro existence is unusual, turns step (2.3);

(2.2) when star sensor quantity is 2, according to the consistance of each star sensor output, judge whether star sensor is unusual, if output is consistent, then star sensor is normal, otherwise star sensor is unusual, turns step (2.3);

(2.3) exist when unusual when gyro or any one parts of star sensor, the consistance of the angular velocity that then records according to gyro and star sensor is determined trouble unit.

Images