CN108613674A - A kind of attitude error suppressing method based on adaptive differential Evolutionary BP neural network - Google Patents
A kind of attitude error suppressing method based on adaptive differential Evolutionary BP neural network Download PDFInfo
- Publication number
- CN108613674A CN108613674A CN201810248952.3A CN201810248952A CN108613674A CN 108613674 A CN108613674 A CN 108613674A CN 201810248952 A CN201810248952 A CN 201810248952A CN 108613674 A CN108613674 A CN 108613674A
- Authority
- CN
- China
- Prior art keywords
- carrier
- moment
- attitude
- navigation
- matrix
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
- G01C21/00—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00
- G01C21/10—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 by using measurements of speed or acceleration
- G01C21/12—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 by using measurements of speed or acceleration executed aboard the object being navigated; Dead reckoning
- G01C21/16—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 by using measurements of speed or acceleration executed aboard the object being navigated; Dead reckoning by integrating acceleration or speed, i.e. inertial navigation
- G01C21/165—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 by using measurements of speed or acceleration executed aboard the object being navigated; Dead reckoning by integrating acceleration or speed, i.e. inertial navigation combined with non-inertial navigation instruments
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
- G01C21/00—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00
- G01C21/02—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 by astronomical means
- G01C21/025—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 by astronomical means with the use of startrackers
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
- G01C21/00—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00
- G01C21/20—Instruments for performing navigational calculations
Landscapes
- Engineering & Computer Science (AREA)
- Radar, Positioning & Navigation (AREA)
- Remote Sensing (AREA)
- Physics & Mathematics (AREA)
- Automation & Control Theory (AREA)
- General Physics & Mathematics (AREA)
- Astronomy & Astrophysics (AREA)
- Navigation (AREA)
Abstract
The invention discloses a kind of attitude error suppressing methods based on adaptive differential Evolutionary BP neural network, belong to field of inertia technology.Including initializing inertia/star sensor integrated navigation system；Acquire the data of star sensor and the output of inertia component；Navigation calculation is carried out to inertia component output data；Posture of the naval vessel relative to inertial system is resolved using star sensor output data；Combining current time using ADE BPNN, interior integrated navigation posture predicts relative inertness system of carrier system attitude angle for the previous period；Star sensor measurement error is compensated using prediction posture and resolves naval vessel position using the star sensor output information after compensation；The navigation information of acquisition is combined navigation calculation；It stores and exports integrated navigation information.The present invention predicts star sensor posture by adaptive differential Evolutionary BP neural network, solves the problems, such as star sensor star chart " hangover " caused by ship sway, the applicability of inertia/star sensor integrated navigation system when enhancing ship sway.
Description
Technical field
The invention belongs to reduce the method for navigation information error in field of inertia technology, and in particular to one kind is based on adaptive
The attitude error suppressing method of differential evolution BP neural network.
Background technology
Inertia/star sensor integrated navigation be it is a kind of with inertial navigation system (Inertial Navigation System,
INS the navigation system based on) can provide real-time speed, posture and location information for carrier.INS mainly utilizes gyroscope and adds
Measurement of velocity loading gage body angular speed and line movable information.Navigation information is obtained after navigation calculation；Star sensor is by shooting star
Scheme and compared with itself star chart library, obtains posture information of the installation carrier relative to inertial space.Inertia/star sensor combination
Navigation system carries out periodical re-graduation on the basis of the navigation information that star sensor resolves, to the divergence expression position error of INS,
Any external information is not needed, is a kind of full autonomous navigation system.Inertia/star sensor integrated navigation system has real-time
By force, the advantages that navigation accuracy is high, complete autonomous, is widely used in marine navigation.However, since naval vessel horizontal pan keeps star quick
The star chart of sensor shooting has " hangover " phenomenon, influences the calculation accuracy of star sensor, and then influences inertia/star sensor combination
Navigation accuracy.
《Chinese inertial technology journal》2010 years volume 18 the 4th interim write by Yang Bo et al.《Missile-borne inertia/satellite/star
Light high-precision integrated navigation》In one text, it is proposed that the online self-calibrating method of gyro error and star sensor installation deviation angle,
Inhibit influence of the device deviation to navigation accuracy；《China's Space science and technology》June the 3rd in 2013 is interim to be write by Zhang Cheng et al.
It writes《Directly re-entry space vehicle inertia/astronomy combined method of sensitivity Horizon》In one text, it is proposed that use star sensor and top
Spiral shell instrument constructs inertial reference, and carries out the astrofix based on infrared horizon on this basis and resolve, and finally carries out inertia/day
The method of literary integrated positioning, it is suppressed that the drift error of gyro；The patent of Publication No. CN105737858A was July 6 in 2016
Disclosed in day《A kind of Airborne Inertial Navigation System attitude parameter calibration method and device》, the high high dynamic star of this method service precision is quick
The dynamic on-line calibration of airborne INS attitude parameters is completed on the basis of the attitude data of sensor output；《IEEE SENSORS
JOURNAL》2017 years volume 17 the 21st interim write by Yang Yanqiang et al.《In-flight Calibration of Gyros
and Star Sensor with Observability Analysis for SINS/CNS Integration》In one text,
By the observability degree for analyzing the error state based on singular value, it is proposed that a kind of four decoupling calibration methods, it is suppressed that device
Influence of the deviation to navigation accuracy, document above are proposed the method for capableing of suppression device deviation, and mostly with missile-borne, airborne
For application background, there is no under environment peculiar to vessel, the star chart " hangover " caused by ship sway this application circumstances
Caused star sensor resolves the inaccurate problem of information and is furtherd investigate.
Invention content
The purpose of the present invention is to provide navigation accuracy is improved, one kind of enhanced navigation information applicability is based on adaptive poor
Divide the attitude error suppressing method of Evolutionary BP neural network.
The purpose of the present invention is realized by following technical solution：
Based on ADE-BPNN, using integrated navigation information in a period of time of inertia/star sensor integrated navigation system as
Input, predicts current t moment attitude of carrier, and attitude of carrier and t moment star sensor are predicted to t moment using Kalman filtering
It resolves attitude of carrier and carries out data fusion, the navigation error of star sensor is estimated, is compensated, obtain star sensor high-precision
Navigation information is combined navigational solution with INS navigation informations.
A kind of attitude error suppressing method based on adaptive differential Evolutionary BP neural network, includes the following steps：
Step 1：It powers on, initializes inertia/star sensor integrated navigation system.
Step 2：Acquire star sensor and inertia component output data.
Step 3：Navigation calculation is carried out to inertia component output data in step 2, obtains carrier positionsCarrier three
To speedCarrier three-axis attitude angleWherein,Latitude, longitude are indicated respectively； East orientation, north orientation, sky orientation speed are indicated respectively；Attitude angle includes pitch angle, roll angle and course angle, respectively by It indicates.
Step 4：By star sensor the output phase in step 2 for the attitude matrix of inertial spaceIt is relatively used to obtain carrier
Property system attitude angleI.e.
Wherein, b indicates that carrier coordinate system, i indicate inertial system,Indicate carrier system to inertial system transition matrix；cbimn(m,
N=1,2,3 it) indicatesIn m rows, the n-th column matrix element；Arcsin is the arcsin function in trigonometric function, and arctan is
Arctan function in trigonometric function；Pitch angle, roll angle and the boat of carrier relative inertness system are indicated respectively
To angle.
Step 5：Utilize ADE-BPNN prediction t moment carrier relative inertnesses system attitude angleWherein, t tables
Show current time,Indicate respectively carrier relative inertness system pitch angle, roll angle and course angle predicted value.
Step 6：Kalman filter 1 is constructed, with star sensor measurement error δ φCi(i=x, y, z) is quantity of state, step
Star sensor resolves the difference that attitude of carrier is predicted in attitude of carrier and step 5 in 4For observed quantity, δ is resolved
φCi(i=x, y, z) compensates step 4 calculation result using calculation result, obtains φCi(i=x, y, z).Wherein, δ
φCx、δφCy、δφCzStar sensor pitch angle, roll angle and course angle error are indicated respectively；φCx、φCy、φCzIt indicates respectively
Pitch angle, roll angle and the course angle of carrier relative inertness system.
Step 7：According to the mathematical relationship between transition matrix, the conversion square that terrestrial coordinate system aligns geographic coordinate system is obtained
Battle arrayTransfer process isIt further resolves and obtains the position of carrier, i.e.,
Wherein,Latitude and longitude are indicated respectively；E indicates that earth system, N indicate quasi- geographic coordinate system,Indicate ground
Transition matrix of the ball system to quasi- geographic coordinate system；ceNmn(m, n=1,2,3) is indicatedIn m rows, the n-th column matrix element；
Transition matrix of the expression carrier system to quasi- geographic coordinate system；Indicate carrier system to inertial system transition matrix；Footmark T indicates square
Battle array transposition；Indicate earth system to inertial system transition matrix.
Step 8：Kalman filter 2 is constructed, the difference that INS in position and step 3 resolves position is resolved with step 7λC-λSFor observed quantity, INS position errorδλS, velocity error δ vSi(i=x, y, z), misalignment Φi(i=x,
Y, z), three axis accelerometer constant error εi(i=x, y, z), three axis accelerometer zero offset error delta Ai(i=x, y, z) is state
Amount resolvesδλS、δvSi(i=x, y, z), Φi(i=x, y, z), εi(i=x, y, z), Δ Ai(i=x, y, z), to step 3
Calculation result compensates, and obtains carrier positionsλ, speed vi(i=x, y, z), attitude angle φi(i=x, y, z).Wherein,δλSLatitude, longitude error are indicated respectively；δvSx、δvSy、δvSzEast orientation, north orientation and sky orientation speed error are indicated respectively；Φx、
Φy、ΦzPitching misalignment, rolling misalignment, course misalignment are indicated respectively.
Step 9：The navigation information carrier positions, speed and the attitude angle that are obtained in step 8 are stored and exported.
The beneficial effects of the present invention are：
After the integrated navigation information for obtaining naval vessel normal work, in conjunction with ADE-BPNN algorithms, it is defeated to predict star sensor
Go out inertial attitude, eliminates the influence that star chart " hangover " exports star sensor attitude angle, navigation calculation is combined with INS, it is defeated
Go out navigation information.(1) applicability for enhancing inertia/star sensor integrated navigation system when horizontal pan occurs in naval vessel, subtracts
The small insufficient problem of the error compensation caused by ship sway；(2) it does not need any extraneous auxiliary information and positioning can be improved
Precision；(3) calculation amount is small, simple easily to realize.
Description of the drawings
Fig. 1 is the Integrated Navigation Algorithm flow chart of the present invention；
Fig. 2 is the Simulation results carried out using the present invention；
Fig. 3 is the actual measurement experimental result carried out using the present invention；
Fig. 4 is actual measurement experiment ship navigation track.
Specific implementation mode
The specific implementation mode of the present invention is described further below in conjunction with the accompanying drawings：
As shown in Figure 1, the present invention provides a kind of Integrated Navigation Algorithm for inertia/star sensor, specifically include as follows
Step：
Step 1：It powers on, initializes inertia/star sensor integrated navigation system.Navigation initialization, need to initialize system：
(1) INS initial values are initialized：The carrier positions of hullλ0(unit degree of being), carrier three-dimensional speed vi0(i=x,
Y, z) (unit is m/s), carrier three-axis attitude angle φi0(i=x, y, z) (unit degree of being)；Initialize transition matrix
Initialize quaternary number q0；
(2) 1 initial parameter values of Kalman filter：State variable initial value δ φCi0(i=x, y, z), mean squared error matrix Pp0, it is
Noise square formation of uniting Qp, measure noise square formation Rp, measure battle array Hp；Remaining initial value is set according to actual conditions.
(3) 2 initial parameter values of Kalman filter：State variable initial value δ vS0i(i=x, y, z),δλS0、Φi0(i=
x,y,z)、εi0(i=x, y, z), Δ Ai0(i=x, y, z), mean squared error matrix P0, system noise square formation Q, measurement noise square formation R,
Measure battle array H；
Remaining initial value is set according to actual conditions.
It is as follows to initialize transform matrix calculations：
Wherein, n indicates navigation system,Indicate the transfer matrix that starting vector system is to navigation；
Initialize quaternary number q0It calculates as follows：
q00=a,
Whereinq0=[q00q01q02q03]T, cbnij(i, j=1,2,3) is indicatedIn the i-th row,
Jth column matrix element；
It in navigation procedure, is updated using the initial information, obtains position, speed and the posture of any time carrier
Angle.
Step 2：System acquires star sensor output matrix in real timeThe 3-axis acceleration of inertia component accelerometer outputWith three axis angular rates of gyroscope outputWherein,Point
Not Biao Shi accelerometer measures specific force in carrier system oxbAxis, oybAxis, ozbComponent on axis (unit is m/s)； Indicate the angular speed of gyroscope measurement in carrier system ox respectivelybAxis, oybAxis, ozb(unit is component on axis
rad/s)。
Step 3：Navigation calculation is carried out to inertia component gathered data in step 2：
Quaternary number attitude matrix updates：
If the rotation quaternary number of any time Relative Navigation system of carrier system is q=[q0q1q2q3]T, the timely amendment of quaternary number
It is as follows：
Wherein,Q is indicated respectivelyi(i=0,1,2,3) change rate；
By the rotation quaternary number q of Relative Navigation system of t moment carrier systemi(t) (i=0,1,2,3) replaces q in formula (4)i(i=
0,1,2,3) change rate of t moment rotation quaternary number, is obtainedIt is in the rotation quaternary number of t+1 moment carriers
Whereinω in formulaiSuperscript b is omitted in (i=x, y, z)；T is the sampling time；As t=1, q (1) is the initial quaternary number q of carrier in step 10。
According to element q in q (t+1)i(t+1) (i=0,1,2,3) updates strap-down matrix
Wherein, the q in above formulai(i=0,1,2,3) it is q in formula (5)i(t+1) (i=0,1,2,3) updates posture information：
Utilize conversion relational expressionAccelerometer is arrived into navigation along what carrier system measured than force information projection transform
System, utilizes following differential equation carrier movement speed：
Wherein, It indicatesChange rate； The specific force of expression accelerometer measures is ox in navigation respectivelynAxis, oynAxis,
oznComponent on axis (unit is m/s)；
The projection f for being in navigation by t moment specific forcei n(t) (i=x, y, z), instead of the f in formula (8)i n(i=x, y, z),
Obtain bearer rate change rateObtain t+1 moment bearer rate and position：
Wherein, R indicates earth radius；As t=1,For
Bearer rate initial value v in step 1i0(i=x, y, z).
So far, bearer rate, position and the attitude angle of INS resolvings are obtained according to formula (7), (9).
Step 4：By star sensor the output phase in step 2 for the attitude matrix of inertial spaceIt is relatively used to obtain carrier
Property system attitude angleI.e.
Step 5：Utilize ADE-BPNN prediction t moment carrier relative inertnesses system attitude angleDetailed process
It is as follows：
(1) initialization operation：Initiation parameter, including population scale N, gene dimension D, mutagenic factor F, crossing rate
CR, and each value range [U of genemin,Umax], utilize formula xij=Umin+rand×(Umax-Umin) random initializtion kind
Group；Wherein i=1,2 ..., N, j=1,2 ..., D, rand indicate to obey equally distributed random number.
(2) judge whether current minimum fitness value reaches μ or iterations reach maximum, if then ADE iteration stoppings,
Go to step (6)；Otherwise it carries out in next step.
(3) offspring subgroup is generated：To each object vectorMutation operation is carried out, variation vector is generatedTo variation vectorIt carries out crossover operation and generates experimental subjectsBy target individualCorresponding experimental subjectsFitness value both is compared in competition, only whenFitness value compared withIt is just chosen as filial generation when more excellent, otherwise directly willAs filial generation.Repeat the production of this step
Raw offspring subgroup.Wherein, G indicates iterations.
(4) adaptive value of assessment offspring subgroup, minimum adaptive value are current optimal values, and corresponding subgroup is current optimal
Individual values.
(5) G=G+1 is enabled, step (2) is gone to.
(6) optimized individual of ADE is assigned as to the initial connection weights and threshold value of BPNN.
(7) by t moment integrated navigation horizontal attitude information φi(i=x, y) and threshold θi(i=x, y) is compared, and works as water
Flat posture information be less than threshold value when judgement naval vessel wave, using t-n~t-1 moment integrated navigation posture informations to BPNN into
Row training, obtains optimal models.
(8) model in (7) is utilized, the attitude angle φ of t moment carrier Relative Navigation system is predictedpi(i=x, y, z) is utilized
Prediction attitude angle obtains carrier system to the transition matrix of navigation systemAccording to the relationship between transition matrix, carrier system phase is obtained
To the transition matrix of inertial systemTransfer process isIt further resolves and obtains carrier relative inertness system
Attitude angle predicted value, i.e.,：
Wherein,It indicatesIn m rows, the n-th column element；Indicate that earth system converts to inertial system
Matrix is obtained by earth rate and navigation time；Indicate that earth system to the transition matrix of navigation system, is combined by the t-1 moment and led
Boat positionλ is obtained.
Step 6：Kalman filter 1 is constructed, with star sensor measurement error δ φCi(i=x, y, z) is quantity of state, step
Star sensor resolves the difference that attitude of carrier is predicted in attitude of carrier and step 5 in 4For observed quantity, specifically
Filtering is as follows：
Using the inertial attitude error model of star sensor, in conjunction with Kalman filtering, error correction is carried out to inertial attitude
It is as follows：
Wherein, Pp(t) indicate that t moment estimates mean square error, Pp(t/t-1) indicate that the t-1 moment is equal to t moment one-step prediction
Square error, Fp(t/t-1) indicate the t-1 moment to the state-transition matrix of t moment, Gp(t-1) the noise driving at t-1 moment is indicated
Battle array, Qp(t-1) t-1 moment system noise variance matrix, H are indicatedpIndicate measurement matrix, RpIndicate measuring noise square difference battle array, Kp(t)
Indicate t moment filtering gain matrix.
T moment state estimation and the update matrix of estimation mean square error are：
Wherein, Xp(t) t moment state estimator, X are indicatedp(t)=[δ φCx δφCy δφCz]T, as t=1, state
Measure Xp(1) element is the quantity of state initial value δ φ of Kalman filter 1 in step 1 inCi0(i=x, y, z).According to calculation result,
Step 4 calculation result is compensated, inertial attitude angle φ is obtainedCi(i=x, y, z).
Step 7：According to the mathematical relationship between transition matrix, the conversion square that terrestrial coordinate system aligns geographic coordinate system is obtained
Battle arrayTransfer process isIt further resolves and obtains the position of carrier, i.e.,
Transition matrix of the carrier system to quasi- Department of GeographyBy the horizontal attitude resolved in step 3It obtains, carries
System is to inertial system transition matrixIt is obtained by calculation result inertial attitude angle in step 6.
Step 8：Kalman filter 2 is constructed, the difference that INS in position and step 3 resolves position is resolved with step 7λC-λSFor observed quantity, INS position errorδλS, velocity error δ vSi(i=x, y, z), misalignment Φi(i=x,
Y, z), three axis accelerometer constant error εi(i=x, y, z), three axis accelerometer zero offset error delta Ai(i=x, y, z) is state
Amount, specific filtering are as follows：
Carrier state amount is missed in conjunction with Kalman filtering using the speed of carrier, position and misalignment error model
Difference is corrected as follows：
Wherein, P (t) indicates that t moment estimates that mean square error, P (t/t-1) indicate that the t-1 moment is square to t moment one-step prediction
Error, F (t/t-1) indicate the t-1 moment to the state-transition matrix of t moment, the noise driving battle array at G (t-1) expression t-1 moment, Q
(t-1) indicate that t-1 moment system noise variance matrix, H indicate measurement matrix, R indicates measuring noise square difference battle array, when K (t) indicates t
Carve filtering gain matrix.
T moment state estimation and the update matrix of estimation mean square error are：
Wherein, X (t) indicates t moment state estimator, and as t=1, element is Kalman in step 1 in quantity of state X (1)
The quantity of state initial value δ v of filter 2S0i(i=x, y, z),δλS0、Φi0(i=x, y, z), εi0(i=x, y, z), Δ Ai0(i
=x, y, z).According to calculation result, INS calculation results in step 3 are compensated, navigation information carrier positions are obtainedλ,
Speed vi(i=x, y, z), attitude angle φi(i=x, y, z).
Step 9：The navigation information carrier positions, speed and the attitude angle that are obtained in step 8 are stored and exported.
Embodiment：
Advantageous effect of the present invention is verified by policy and actual measurement：
Emulation experiment：
Ship's navigation route：Linear motion
System initialization parameter is as follows：
Ship Motion parameter：Ship velocity：vx0=0m/s, vy0=0m/s, vz0=0m/s
Ship gesture：
Wave the influence to ship velocity：
It is evenly distributed in [0,2 π]
High-frequency vibrates the influence to ship velocity：
It is evenly distributed in [0,2 π]
Local longitude and latitude：λ0=126.6705 °
Gyro parameter：Gyroscope constant value drift：0.01°/h
Gyro scale factor error：1×10-5
Gyro white noise error：0.001°/h
Accelerometer parameter：Accelerometer bias：10-4g
Accelerometer scale factor error：1×10-5
Accelerometer white noise error：10-5g
ADE-BPNN parameters：
Best fit network has 20 input neurons, 10 hidden neurons and 1 output neuron
Maximum iteration：100 times
Training objective error：0.00004
Learning rate：0.1
Hide layer functions：logsig
Activate layer functions：purelin
Training function：trainlm
Connection weight and threshold range：[-1,1]
Population size：N=50
Maximum iteration：GenM=50
Required precision：μ=0.15
Crossing-over rate：CR=0.1
Mutation factor minimum value：Fmin=0.2
Mutation factor maximum value：Fmax=0.6
1 numerical value of Kalman filter is arranged：
Pp0=diag ([0.01 0.01 0.01]2)
Qp=diag ([0.01 0.01 0.01]2)
Rp=diag ([0.001 0.001 0.001]2)
2 numerical value of Kalman filter is arranged：
P0=diag ([10 10 0.0005 0.0005 0.05 0.05 0.15 0.01 0.01 0.01 10-5g 10-5g
10-5g]2)
Q=diag ([0.5 × 10-5g 0.5×10-5g O1×2 0.5×0.01 0.5×0.01 0.5×0.01 O1×6
]2)
R=diag ([100R 100R]2)
Sample frequency：0.1s
Sampling time：1h
Actual measurement experiment：
Ship navigation route：As shown in Figure 4.
Gyro parameter：Dynamic range：±100°/s
Zero bias：≤0.01°/h
The nonlinearity of scale factor：≤20ppm
Star sensor parameter：Field-of-view angle：24°
Interspace observation is horizontal：Not less than 7 grades
Data update frequency：20Hz
Attitude accuracy：5″
Using invention the method, obtain sensitive in the hull inertia for having stormy waves with and without ADE-BPNN algorithms/star
Device integrated navigation information error curve.Fig. 2 indicates that simulation result comparison curves, Fig. 3 expressions are actual measurement Comparison of experiment results curves.
The result shows that present invention decreases when hull horizontal pan, star chart " hangover " resolves star sensor the influence of navigation information, into
And inertia/star sensor integrated navigation error is preferably inhibited, actual demand can be met.
The foregoing is only a preferred embodiment of the present invention, is not intended to restrict the invention, for the skill of this field
For art personnel, the invention may be variously modified and varied.All within the spirits and principles of the present invention, any made by repair
Change, equivalent replacement, improvement etc., should all be included in the protection scope of the present invention.
Claims (8)
1. a kind of attitude error suppressing method based on adaptive differential Evolutionary BP neural network, which is characterized in that including following
Step：
(1) it powers on, initializes inertia/star sensor integrated navigation system；
(2) system acquires star sensor output matrix in real timeThe 3-axis acceleration of inertia component accelerometer outputWith three axis angular rates of gyroscope outputWherein,Point
Not Biao Shi accelerometer measures specific force in carrier system oxbAxis, oybAxis, ozbComponent on axis, unit are m/s； Indicate the angular speed of gyroscope measurement in carrier system ox respectivelybAxis, oybAxis, ozbComponent on axis, unit are rad/s；
(3) navigation calculation is carried out to inertia component gathered data in step (2)；
(4) by star sensor the output phase in step (2) for the attitude matrix of inertial spaceObtain carrier relative inertness system appearance
State angle
(5) ADE-BPNN prediction t moment carrier relative inertnesses system attitude angle is utilizedWherein, when t indicates current
It carves,Indicate respectively carrier relative inertness system pitch angle, roll angle and course angle predicted value；
(6) Kalman filter 1 is constructed, with star sensor measurement error δ φCi(i=x, y, z) is quantity of state, step (4) culminant star
Sensor resolves the difference of prediction attitude of carrier in attitude of carrier and step (5)For observed quantity；According to resolving
As a result, being compensated to step (4) calculation result, inertial attitude angle φ is obtainedCi(i=x, y, z)；
(7) according to the mathematical relationship between transition matrix, the transition matrix that terrestrial coordinate system aligns geographic coordinate system is obtainedTurn
The process of changing isIt further resolves and obtains the position of carrier, transition matrix of the carrier system to quasi- Department of GeographyBy the horizontal attitude resolved in step (3)It obtains, carrier system to inertial system transition matrixBy step (6)
Middle calculation result inertial attitude angle obtains；
(8) Kalman filter 2 is constructed, the difference that INS in position and step (3) resolves position is resolved with step (7)λC-
λSFor observed quantity, INS position errorVelocity error δ vSi(i=x, y, z), misalignment Φi(i=x, y, z), three axis
Gyroscope constant value error εi(i=x, y, z), three axis accelerometer zero offset error delta Ai(i=x, y, z) is quantity of state, according to solution
It calculates as a result, compensated to INS calculation results in step (3), obtains navigation information carrier positionsλ, speed vi(i=x, y,
Z), attitude angle φi(i=x, y, z)；
(9) the navigation information carrier positions, speed and the attitude angle that are obtained in step (8) are stored and is exported.
2. a kind of attitude error suppressing method based on adaptive differential Evolutionary BP neural network according to claim 1,
It is characterized in that, the step (1) specifically includes：
(1.1) INS initial values are initialized：The carrier positions of hullλ0(unit degree of being), carrier three-dimensional speed vi0(i=x, y,
Z) (unit is m/s), carrier three-axis attitude angle φi0(i=x, y, z) (unit degree of being)；Initialize transition matrixJust
Beginningization quaternary number q0；
(1.2) 1 initial parameter values of Kalman filter：State variable initial value δ φCi0(i=x, y, z), mean squared error matrix Pp0, system
Noise square formation Qp, measure noise square formation Rp, measure battle array Hp；Remaining initial value is set according to actual conditions.
(1.3) 2 initial parameter values of Kalman filter：State variable initial value δ vS0i(i=x, y, z),δλS0、Φi0(i=x,
y,z)、εi0(i=x, y, z), Δ Ai0(i=x, y, z), mean squared error matrix P0, system noise square formation Q, measurement noise square formation R, amount
Survey battle array H；
Remaining initial value is set according to actual conditions.
It is as follows to initialize transform matrix calculations：
Wherein, n indicates navigation system,Indicate the transfer matrix that starting vector system is to navigation；
Initialize quaternary number q0It calculates as follows：
q00=a,Whereinq0=[q00 q01 q02 q03]T, cbnij(i, j=1,2,3) is indicatedIn the i-th row,
J column matrix elements；
It in navigation procedure, is updated using the initial information, obtains position, speed and the attitude angle of any time carrier.
3. a kind of attitude error suppressing method based on adaptive differential Evolutionary BP neural network according to claim 1,
It is characterized in that, the step (3) specifically includes：
(3.1) quaternary number attitude matrix updates：
If the rotation quaternary number of any time Relative Navigation system of carrier system is q=[q0 q1 q2 q3]T, the timely amendment of quaternary number
It is as follows：
Wherein,Q is indicated respectivelyi(i=0,1,2,3) change rate；
(3.2) by the rotation quaternary number q of Relative Navigation system of t moment carrier systemi(t) (i=0,1,2,3) replaces q in formula (4)i(i=
0,1,2,3) change rate of t moment rotation quaternary number, is obtainedIt is in the rotation quaternary number of t+1 moment carriers
Whereinω in formulaiSuperscript b is omitted in (i=x, y, z)；T is the sampling time；As t=1, q (1) is the initial quaternary number q of carrier in step 10。
(3.3) according to element q in q (t+1)i(t+1) (i=0,1,2,3) updates strap-down matrix
Wherein, the q in above formulai(i=0,1,2,3) it is q in formula (5)i(t+1) (i=0,1,2,3) updates posture information：
(3.4) conversion relational expression is utilizedAccelerometer is arrived into navigation along what carrier system measured than force information projection transform
System, utilizes following differential equation carrier movement speed：
Wherein,It indicatesChange rate； The specific force of expression accelerometer measures is ox in navigation respectivelynAxis, oynAxis, ozn(unit is equal for component on axis
For m/s)；
(3.5) the projection f for being in navigation by t moment specific forcei n(t) (i=x, y, z), instead of the f in formula (8)i n(i=x, y, z),
Obtain bearer rate change rateObtain t+1 moment bearer rate and position：
Wherein, R indicates earth radius；As t=1,For step
Bearer rate initial value v in 1i0(i=x, y, z)；
So far, bearer rate, position and the attitude angle of INS resolvings are obtained according to formula (5), (7).
4. a kind of attitude error suppressing method based on adaptive differential Evolutionary BP neural network according to claim 1,
It is characterized in that, the step (4) specifically includes：
Carrier relative inertness system attitude angleFor：
5. a kind of attitude error suppressing method based on adaptive differential Evolutionary BP neural network according to claim 1,
It is characterized in that, the step (6) specifically includes：
(6.1) the inertial attitude error model for utilizing star sensor carries out error correction in conjunction with Kalman filtering to inertial attitude
It is as follows：
Wherein, Pp(t) indicate that t moment estimates mean square error, Pp(t/t-1) indicate the t-1 moment to t moment one-step prediction mean square error
Difference, Fp(t/t-1) indicate the t-1 moment to the state-transition matrix of t moment, Gp(t-1) indicate that the noise at t-1 moment drives battle array, Qp
(t-1) t-1 moment system noise variance matrix, H are indicatedpIndicate measurement matrix, RpIndicate measuring noise square difference battle array, Kp(t) t is indicated
Moment filtering gain matrix；
(6.2) the update matrix of t moment state estimation and estimation mean square error is：
Wherein, Xp(t) t moment state estimator, X are indicatedp(t)=[δ φCx δφCy δφCz]T, as t=1, quantity of state Xp
(1) element is the quantity of state initial value δ φ of Kalman filter 1 in step 1 inCi0(i=x, y, z), according to calculation result, to step
Suddenly (4) calculation result compensates, and obtains inertial attitude angle φCi(i=x, y, z).
6. a kind of attitude error suppressing method based on adaptive differential Evolutionary BP neural network according to claim 1,
It is characterized in that, the step (7) specifically includes：
According to the mathematical relationship between transition matrix, the transition matrix that terrestrial coordinate system aligns geographic coordinate system is obtainedConversion
Process isIt further resolves and obtains the position of carrier, i.e.,
Transition matrix of the carrier system to quasi- Department of GeographyBy the horizontal attitude resolved in step (3)It obtains, carrier
It is to inertial system transition matrixIt is obtained by calculation result inertial attitude angle in step (6).
7. a kind of attitude error suppressing method based on adaptive differential Evolutionary BP neural network according to claim 1,
It is characterized in that, the step (8) specifically includes：
Using the speed of carrier, position and misalignment error model, in conjunction with Kalman filtering, error is carried out to carrier state amount and is repaiied
As under：
Wherein, P (t) indicates that t moment estimates that mean square error, P (t/t-1) indicate the t-1 moment to t moment one-step prediction mean square error
Difference, F (t/t-1) indicate the t-1 moment to the state-transition matrix of t moment, the noise driving battle array at G (t-1) expression t-1 moment, Q
(t-1) indicate that t-1 moment system noise variance matrix, H indicate measurement matrix, R indicates measuring noise square difference battle array, when K (t) indicates t
Carve filtering gain matrix；
T moment state estimation and the update matrix of estimation mean square error are：
Wherein, X (t) indicates t moment state estimator, and as t=1, element is Kalman filtering in step 1 in quantity of state X (1)
The quantity of state initial value δ v of device 2S0i(i=x, y, z),δλS0、Φi0(i=x, y, z), εi0(i=x, y, z), Δ Ai0(i=
x,y,z).According to calculation result, INS calculation results in step 3 are compensated, navigation information carrier positions are obtainedλ, speed
vi(i=x, y, z), attitude angle φi(i=x, y, z).
8. a kind of attitude error suppressing method based on adaptive differential Evolutionary BP neural network according to claim 1,
It is characterized in that, the step (5) specifically includes：
(5.1) initialization operation：Initiation parameter, including population scale N, gene dimension D, mutagenic factor F, crossing rate CR,
And value range [the U of each genemin,Umax], utilize formula xij=Umin+rand×(Umax-Umin) random initializtion population；
Wherein i=1,2 ..., N, j=1,2 ..., D, rand indicate to obey equally distributed random number；
(5.2) judge whether current minimum fitness value reaches μ or iterations reach maximum, if then ADE iteration stoppings, turns
To step (5.6)；Otherwise it carries out in next step；
(5.3) offspring subgroup is generated：To each object vectorMutation operation is carried out, variation vector is generatedTo variation vectorIt carries out crossover operation and generates experimental subjectsBy target individualCorresponding experimental subjectsFitness value both is compared in competition, only whenFitness value compared withIt is just chosen as filial generation when more excellent, otherwise directly willAs filial generation.Repeat the production of this step
Raw offspring subgroup, wherein G indicates iterations；
(5.4) adaptive value of assessment offspring subgroup, minimum adaptive value are current optimal values, and corresponding subgroup is current optimal
Body value；
(5.5) G=G+1 is enabled, step (5.2) is gone to；
(5.6) optimized individual of ADE is assigned as to the initial connection weights and threshold value of BPNN；
(5.7) by t moment integrated navigation horizontal attitude information φi(i=x, y) and threshold θi(i=x, y) is compared, and works as level
Judgement naval vessel waves when posture information is less than threshold value, is carried out to BPNN using t-n~t-1 moment integrated navigation posture informations
Training, obtains optimal models；
(5.8) model obtained in step (5.7) is utilized, the attitude angle φ of t moment carrier Relative Navigation system is predictedpi(i=x,
Y, z), obtain the transition matrix that carrier system is to navigation using prediction attitude angleAccording to the relationship between transition matrix, obtain
To the transition matrix of relative inertness system of carrier systemTransfer process isIt further resolves and obtains carrier
The attitude angle predicted value of relative inertness system, i.e.,：
Wherein,It indicatesIn m rows, the n-th column element；Earth system is indicated to inertial system transition matrix,
It is obtained by earth rate and navigation time；The transition matrix that earth system is to navigation is indicated, by t-1 moment integrated navigations positionλ is obtained.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810248952.3A CN108613674A (en) | 2018-03-25 | 2018-03-25 | A kind of attitude error suppressing method based on adaptive differential Evolutionary BP neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810248952.3A CN108613674A (en) | 2018-03-25 | 2018-03-25 | A kind of attitude error suppressing method based on adaptive differential Evolutionary BP neural network |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108613674A true CN108613674A (en) | 2018-10-02 |
Family
ID=63658740
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810248952.3A Pending CN108613674A (en) | 2018-03-25 | 2018-03-25 | A kind of attitude error suppressing method based on adaptive differential Evolutionary BP neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108613674A (en) |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109579853A (en) * | 2019-01-24 | 2019-04-05 | 燕山大学 | Inertial navigation indoor orientation method based on BP neural network |
| CN109737955A (en) * | 2018-12-14 | 2019-05-10 | 南京理工大学 | A kind of attitude prediction method of wave compensation system |
| CN110118979A (en) * | 2018-11-26 | 2019-08-13 | 太原理工大学 | The method of improved differential evolution algorithm estimation multipath parameter based on broad sense cross-entropy |
| CN110898409A (en) * | 2019-11-05 | 2020-03-24 | 五邑大学 | Intelligent badminton racket action recognition system |
| CN110986936A (en) * | 2019-12-17 | 2020-04-10 | 武汉理工大学 | Passenger ship personnel positioning and navigation method based on edge calculation |
| CN111027137A (en) * | 2019-12-05 | 2020-04-17 | 中国人民解放军63620部队 | High-precision dynamic construction method of spacecraft dynamics model based on telemetering data |
| CN111637882A (en) * | 2020-05-23 | 2020-09-08 | 西北工业大学 | Differential evolution geomagnetic navigation method based on grid features |
| CN112291676A (en) * | 2020-05-18 | 2021-01-29 | 珠海市杰理科技股份有限公司 | Method and system for inhibiting audio signal tailing, chip and electronic equipment |
| US11168984B2 (en) * | 2019-02-08 | 2021-11-09 | The Boeing Company | Celestial navigation system and method |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20020004691A1 (en) * | 2000-03-10 | 2002-01-10 | Yasuhiro Kinashi | Attitude determination and alignment using electro-optical sensors and global navigation satellites |
| CN102445200A (en) * | 2011-09-30 | 2012-05-09 | 南京理工大学 | Microminiature personal combined navigation system as well as navigating and positioning method thereof |
| CN102519470A (en) * | 2011-12-09 | 2012-06-27 | 南京航空航天大学 | Multi-level embedded integrated navigation system and navigation method |
| CN103913180A (en) * | 2014-03-26 | 2014-07-09 | 中国科学院长春光学精密机械与物理研究所 | Mounting angle calibration method for onboard large-view-field high-precision star sensor |
| CN104034329A (en) * | 2014-06-04 | 2014-09-10 | 南京航空航天大学 | Multi-integrated navigation processing device under launch inertial system and navigation method of multi-integrated navigation processing device |
-
2018
- 2018-03-25 CN CN201810248952.3A patent/CN108613674A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20020004691A1 (en) * | 2000-03-10 | 2002-01-10 | Yasuhiro Kinashi | Attitude determination and alignment using electro-optical sensors and global navigation satellites |
| CN102445200A (en) * | 2011-09-30 | 2012-05-09 | 南京理工大学 | Microminiature personal combined navigation system as well as navigating and positioning method thereof |
| CN102519470A (en) * | 2011-12-09 | 2012-06-27 | 南京航空航天大学 | Multi-level embedded integrated navigation system and navigation method |
| CN103913180A (en) * | 2014-03-26 | 2014-07-09 | 中国科学院长春光学精密机械与物理研究所 | Mounting angle calibration method for onboard large-view-field high-precision star sensor |
| CN104034329A (en) * | 2014-06-04 | 2014-09-10 | 南京航空航天大学 | Multi-integrated navigation processing device under launch inertial system and navigation method of multi-integrated navigation processing device |
Non-Patent Citations (1)
| Title |
|---|
| 齐昭: "舰用捷联惯导/天文/计程仪组合导航关键技术研究", 《中国优秀博士学位论文全文数据库•工程科技Ⅱ辑》 * |
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110118979A (en) * | 2018-11-26 | 2019-08-13 | 太原理工大学 | The method of improved differential evolution algorithm estimation multipath parameter based on broad sense cross-entropy |
| CN110118979B (en) * | 2018-11-26 | 2023-02-28 | 太原理工大学 | Method for estimating multipath parameters by using improved differential evolution algorithm based on generalized mutual entropy |
| CN109737955A (en) * | 2018-12-14 | 2019-05-10 | 南京理工大学 | A kind of attitude prediction method of wave compensation system |
| CN109579853A (en) * | 2019-01-24 | 2019-04-05 | 燕山大学 | Inertial navigation indoor orientation method based on BP neural network |
| US11168984B2 (en) * | 2019-02-08 | 2021-11-09 | The Boeing Company | Celestial navigation system and method |
| CN110898409A (en) * | 2019-11-05 | 2020-03-24 | 五邑大学 | Intelligent badminton racket action recognition system |
| CN111027137A (en) * | 2019-12-05 | 2020-04-17 | 中国人民解放军63620部队 | High-precision dynamic construction method of spacecraft dynamics model based on telemetering data |
| CN111027137B (en) * | 2019-12-05 | 2023-07-14 | 中国人民解放军63620部队 | High-precision dynamic construction method for spacecraft dynamics model based on telemetry data |
| CN110986936A (en) * | 2019-12-17 | 2020-04-10 | 武汉理工大学 | Passenger ship personnel positioning and navigation method based on edge calculation |
| CN112291676A (en) * | 2020-05-18 | 2021-01-29 | 珠海市杰理科技股份有限公司 | Method and system for inhibiting audio signal tailing, chip and electronic equipment |
| CN112291676B (en) * | 2020-05-18 | 2021-10-15 | 珠海市杰理科技股份有限公司 | Method and system for inhibiting audio signal tailing, chip and electronic equipment |
| CN111637882A (en) * | 2020-05-23 | 2020-09-08 | 西北工业大学 | Differential evolution geomagnetic navigation method based on grid features |
| CN111637882B (en) * | 2020-05-23 | 2022-11-11 | 西北工业大学 | Differential evolution geomagnetic navigation method based on grid features |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108613674A (en) | A kind of attitude error suppressing method based on adaptive differential Evolutionary BP neural network | |
| CN108827310B (en) | Marine star sensor auxiliary gyroscope online calibration method | |
| Fang et al. | Predictive iterated Kalman filter for INS/GPS integration and its application to SAR motion compensation | |
| CN109813311B (en) | Unmanned aerial vehicle formation collaborative navigation method | |
| WO2020062807A1 (en) | Method for application of improved unscented kalman filter algorithm in underwater integrated navigation | |
| CN106500693B (en) | A kind of AHRS algorithm based on adaptive extended kalman filtering | |
| CN104655131B (en) | Inertial navigation Initial Alignment Method based on ISTSSRCKF | |
| CN105737823B (en) | A kind of GPS/SINS/CNS Combinated navigation methods based on five rank CKF | |
| CN102829777B (en) | Autonomous underwater vehicle combined navigation system and method | |
| CN103744098B (en) | AUV integrated navigation systems based on SINS/DVL/GPS | |
| CN106980133A (en) | The GPS INS Combinated navigation methods and system for being compensated and being corrected using neural network algorithm | |
| CN108226980A (en) | Difference GNSS and the adaptive close coupling air navigation aids of INS based on Inertial Measurement Unit | |
| CN104697520B (en) | Integrated gyro free strap down inertial navigation system and gps system Combinated navigation method | |
| Pan et al. | Underwater Doppler navigation with self-calibration | |
| CN112146655B (en) | Elastic model design method for BeiDou/SINS tight integrated navigation system | |
| CN106441291B (en) | A kind of integrated navigation system and air navigation aid based on strong tracking SDRE filtering | |
| CN106017460B (en) | A kind of underwater hiding-machine navigation locating method of terrain aided inertial navigation tight integration | |
| CN108548542A (en) | A kind of LEO based on atmospheric drag acceleration analysis determines method | |
| CN109708663B (en) | Star sensor online calibration method based on aerospace plane SINS assistance | |
| CN109724624A (en) | A kind of airborne adaptive Transfer alignment algorithm suitable for wing flexure deformation | |
| CN108592943A (en) | A kind of inertial system coarse alignment computational methods based on OPREQ methods | |
| CN106767928A (en) | A kind of self adaptation fast transfer alignment method | |
| CN113503892A (en) | Inertial navigation system moving base initial alignment method based on odometer and backtracking navigation | |
| CN116105730A (en) | Angle measurement-only optical combination navigation method based on cooperative target satellite very short arc observation | |
| CN115265532A (en) | Auxiliary filtering method for marine integrated navigation |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20181002 |
|
| RJ01 | Rejection of invention patent application after publication |