CN113447019B - INS/CNS integrated navigation method, system, storage medium and equipment - Google Patents

INS/CNS integrated navigation method, system, storage medium and equipment Download PDF

Info

Publication number
CN113447019B
CN113447019B CN202110898503.5A CN202110898503A CN113447019B CN 113447019 B CN113447019 B CN 113447019B CN 202110898503 A CN202110898503 A CN 202110898503A CN 113447019 B CN113447019 B CN 113447019B
Authority
CN
China
Prior art keywords
error
ins
navigation system
subsystem
attitude
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.)
Active
Application number
CN202110898503.5A
Other languages
Chinese (zh)
Other versions
CN113447019A (en
Inventor
柳笛
李庆华
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Qilu University of Technology
Original Assignee
Qilu University of Technology
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Qilu University of Technology filed Critical Qilu University of Technology
Priority to CN202110898503.5A priority Critical patent/CN113447019B/en
Publication of CN113447019A publication Critical patent/CN113447019A/en
Application granted granted Critical
Publication of CN113447019B publication Critical patent/CN113447019B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

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
    • G01C21/20—Instruments for performing navigational calculations

Landscapes

  • Engineering & Computer Science (AREA)
  • Radar, Positioning & Navigation (AREA)
  • Remote Sensing (AREA)
  • Automation & Control Theory (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Navigation (AREA)

Abstract

The invention relates to an INS/CNS integrated navigation method, a system, a storage medium and equipment, wherein the INS/CNS integrated navigation method comprises the following steps: acquiring attitude, speed and position information of a target, and acquiring a non-Gaussian measurement noise variance and a predicted value of an optimal estimation error in the attitude, speed and position information based on a trained neural network; correcting the optimal estimation error value in the INS/CNS integrated navigation system based on the non-Gaussian measured noise variance and the predicted value of the optimal estimation error; and correcting the output target attitude, speed and position information based on the corrected optimal estimation error value, thereby obtaining the navigation information output by the INS/CNS integrated navigation system. The problem of low navigation precision of the INS/CNS integrated navigation system in a non-Gaussian measurement noise environment is solved, and the INS/CNS integrated navigation system is suitable for the non-Gaussian measurement noise environment.

Description

INS/CNS integrated navigation method, system, storage medium and device
Technical Field
The invention relates to the field of integrated navigation control, in particular to an INS/CNS integrated navigation method, system, storage medium and equipment.
Background
The statements in this section merely provide background information related to the present disclosure and may not necessarily constitute prior art.
The INS/CNS integrated navigation system can realize high-precision navigation positioning, and the traditional Kalman filtering algorithm involved in the data processing process in the navigation process realizes the optimal estimation of state quantity through minimum mean square error estimation under the condition of supposing that an integrated navigation system model meets Gaussian distribution. However, due to the complex motion environment of the carrier/target where the INS/CNS integrated navigation system is located, the INS (inertial navigation system) and CNS (astronomical navigation system) measurements of the INS/CNS integrated navigation system are usually affected by non-gaussian noise, and if a filtering algorithm based on gaussian assumption is still used in this situation, performance of the filter may be degraded, and even a phenomenon of filter divergence may occur, so that the navigation accuracy of the INS/CNS integrated navigation system may be seriously degraded.
Disclosure of Invention
In order to solve at least one technical problem in the background art, the present invention provides an INS/CNS integrated navigation method, system, storage medium and device, which overcome the problems of reduced navigation accuracy of the INS/CNS integrated navigation system and low execution efficiency of the conventional INS/CNS integrated navigation method in the non-gaussian measurement noise environment.
In order to achieve the purpose, the invention adopts the following technical scheme:
a first aspect of the invention provides a method of INS/CNS integrated navigation comprising the steps of:
acquiring attitude, speed and position information of a target, and acquiring a predicted value of non-Gaussian measurement noise variance and an optimal estimation error in the attitude, speed and position information based on a trained neural network;
correcting the optimal estimation error value in the INS/CNS integrated navigation system based on the non-Gaussian measured noise variance and the predicted value of the optimal estimation error;
and correcting the output target attitude, speed and position information based on the corrected optimal estimation error value, thereby obtaining the navigation information output by the INS/CNS integrated navigation system.
The INS/CNS integrated navigation system has a first subsystem and a second subsystem.
The training process of the neural network comprises the steps that the first subsystem takes a platform misalignment angle, a speed error, a position error, gyro constant drift and accelerometer constant bias as state quantities, and takes the difference between attitude, position and apparent height data output by an astronomical navigation system and attitude, position and apparent height data output by an inertial navigation system as an observed quantity to construct an INS/CNS integrated navigation system model; based on an INS/CNS integrated navigation system model, the Kalman filtering algorithm of the maximum correlation entropy is utilized to obtain the optimal filtering estimation values of attitude error, speed error and position error in the output data of the inertial navigation system.
The training process of the neural network further comprises the steps of training the neural network of the first subsystem by taking the filter gain sequence obtained at the first n moments in the filtering process and the measurement noise variance matrix obtained in the filtering process at the current moment as input and output samples respectively, and outputting an optimal learning training result, wherein n is more than or equal to 1.
The training process of the neural network further comprises the steps that the second subsystem takes a platform misalignment angle, a speed error, a position error, gyro constant drift and accelerometer constant bias as state quantities, the difference between attitude, position and apparent height data output by the astronomical navigation system and attitude, position and apparent height data output by the inertial navigation system is taken as system view measurement, the optimal training result of the neural network of the first subsystem is input into a volume Kalman filter of the second subsystem, optimal filtering estimation is carried out on the error of the inertial navigation system, and the optimal values of the attitude error, the speed error and the position error are obtained.
The training process of the neural network further comprises the steps of obtaining the difference values of the attitude error, the speed error and the position error of the first subsystem and the second subsystem, respectively serving as the input sample and the output sample of the neural network of the second subsystem for training based on the difference values and the filtering gain of the current moment obtained by the second subsystem, and outputting the optimal learning training result.
Under the condition that the measurement noise is non-Gaussian noise, the filter gain sequence of the first n moments in the second subsystem is input into the trained neural network of the first subsystem, and a variance matrix of the measurement noise at the current moment is predicted, wherein n is more than or equal to 1; the second subsystem obtains a filtering estimation value of a system error based on a variance matrix of the measured noise at the current moment, obtains a filtering gain at the current moment, takes the filtering gain as the input of a neural network of the second subsystem, and predicts an error value of optimal estimation at the current moment; and correcting the system error of the inertial navigation system based on the predicted error value to obtain a corrected attitude error, a corrected speed error and a corrected position error.
A second aspect of the invention provides an INS/CNS combination navigation system comprising:
a prediction module configured to: acquiring attitude, speed and position information of a target, and acquiring a predicted value of non-Gaussian measurement noise variance and an optimal estimation error in the attitude, speed and position information based on a trained neural network;
a compensation module configured to: correcting the optimal estimation error value in the INS/CNS integrated navigation system based on the non-Gaussian measured noise variance and the predicted value of the optimal estimation error;
an output module configured to: and correcting the output target attitude, speed and position information based on the corrected optimal estimation error value, thereby obtaining the navigation information output by the INS/CNS integrated navigation system.
A third aspect of the invention provides a computer-readable storage medium.
A computer readable storage medium having stored thereon a computer program which, when executed by a processor, carries out the steps of the INS/CNS combination navigation method as described above.
A fourth aspect of the invention provides a computer apparatus.
A computer apparatus comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the processor when executing the program implementing the steps in the combined INS/CNS navigation method as described above.
Compared with the prior art, the above one or more technical schemes have the following beneficial effects:
1. the problem of low navigation precision of the INS/CNS integrated navigation system in a non-Gaussian measurement noise environment is solved, and the INS/CNS integrated navigation system is suitable for the non-Gaussian measurement noise environment.
2. Compared with an INS/CNS integrated navigation algorithm based on an information theory optimization criterion, the execution efficiency of the navigation process is obviously improved.
3. On the premise of not increasing the hardware cost and the like, the precision and the real-time performance of the INS/CNS integrated navigation system are improved.
Drawings
The accompanying drawings, which are incorporated in and constitute a part of this specification, are included to provide a further understanding of the invention, and are incorporated in and constitute a part of this specification, illustrate exemplary embodiments of the invention and together with the description serve to explain the invention and not to limit the invention.
FIG. 1 is a schematic diagram of an INS/CNS integrated navigation system provided by one or more embodiments of the present invention in an initial training phase;
fig. 2 is a schematic diagram of an INS/CNS integrated navigation system in real-time estimation and error compensation of measured noise variance according to one or more embodiments of the present invention.
Detailed Description
The invention is further described with reference to the following figures and examples.
It is to be understood that the following detailed description is exemplary and is intended to provide further explanation of the invention as claimed. Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.
It is noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of exemplary embodiments according to the invention. As used herein, the singular forms "a", "an" and "the" are intended to include the plural forms as well, and it should be understood that when the terms "comprises" and/or "comprising" are used in this specification, they specify the presence of stated features, steps, operations, devices, components, and/or combinations thereof, unless the context clearly indicates otherwise.
The INS/CNS integrated navigation system is a navigation system formed by assisting a star sensor on the basis of a traditional pure inertial navigation system. The INS is an autonomous navigation system which uses an accelerometer to measure the motion acceleration of a carrier under a reference coordinate system (inertial coordinate system) established by a gyroscope to calculate the position, speed and attitude of the carrier in real time. Usually consisting of an inertial measurement unit and a processing unit. The inertial measurement unit comprises an accelerometer and a gyroscope, the gyroscope is used for measuring the rotation motion of the carrier, and the accelerometer is used for measuring the translation motion acceleration of the carrier; and the processing unit calculates the speed, position and attitude data of the carrier according to the measured acceleration signal and angular motion signal, and finishes the output of the carrier navigation information. The navigation system completes the navigation task independently by completely depending on an inertial device, does not contact with the outside, and has the advantages of high short-time precision, continuous output, strong anti-interference capability, capability of providing position and attitude information and the like; however, the navigation error is accumulated with time, and the error of long-time work is large. The CNS is a navigation system for calculating the position of a carrier by using the direction information of a celestial body measured by a celestial body sensor. The celestial body is used as a navigation information source, so that the celestial body is good in concealment and strong in autonomy, not only can provide position information, but also can provide high-precision attitude information, and output information is discontinuous and is easily influenced by the environment. Because inertia and astronomical navigation systems have advantages and disadvantages, the inertia and astronomical navigation systems are combined to form an INS/CNS integrated navigation system, errors such as attitude errors and gyro constant drift are corrected by using starlight information, advantage complementation is realized, the precision and reliability of the navigation system are improved, and the inertial navigation system and the astronomical navigation system become an important development direction of guided missile and airplane navigation technologies.
As described in the background art, due to the complex motion environment of the carrier/target on which the INS/CNS integrated navigation system is located, the INS (inertial navigation system) and CNS (astronomical navigation system) measurements of the INS/CNS integrated navigation system are usually affected by non-gaussian noise, and if a filtering algorithm based on the gaussian assumption is still used in such a situation, performance degradation of the filter and even a phenomenon of filter divergence occur, which leads to a serious reduction in the navigation accuracy of the INS/CNS integrated navigation system.
The optimization criterion in the information theory uses the information argument (such as entropy) estimated from the data to replace the commonly used second-order statistics such as variance and covariance as the optimization index. Especially the maximum correlation entropy criterion in the information theory, it can realize the statistical analysis of the second order and above error terms. Therefore, kalman filtering algorithms based on maximum correlation entropy are proposed and used in large quantities (e.g. the MCGHCKF algorithm). However, some parameters in such algorithms are set based on human experience, and the metrology update process is inefficient due to the presence of iterative optimization.
Therefore, in order to better adapt the INS/CNS integrated navigation algorithm to engineering applications and to be applicable to non-gaussian measured noise environments, the following embodiments provide an INS/CNS integrated navigation method, system, storage medium and device applicable to non-gaussian measured noise environments, and the INS/CNS integrated navigation system is trained by using the neural network and using the attitude error, position error and refractive vision height error of the INS and the CNS as observed quantities, so that the INS/CNS integrated navigation system has a function of predicting the measured noise variance and the optimal estimation error value; under the condition that the measured noise is non-Gaussian noise, a variance matrix of the measured noise is predicted in real time by using a neural network, the optimal estimation error value of the CKF is predicted and corrected based on the predicted measured variance matrix, the carrier/target attitude, speed and position information output by the INS are corrected based on the corrected optimal estimation value, and finally the INS/CNS integrated navigation system can output high-precision navigation information under the non-Gaussian noise measuring environment.
The first embodiment is as follows:
the present embodiment aims to provide an INS/CNS integrated navigation method, comprising the following steps:
(1) In an initial training stage, a maximum correlation entropy generalized high-order volume Kalman filter (MCGHCKF) in the subsystem 1 and a volume Kalman filter (CKF) in the subsystem 2 respectively finish training of an LSTM neural network contained in the same subsystem by taking attitude errors, position errors and refraction visual height errors of INS and CNS as observed quantities, so that the LSTM neural networks respectively have prediction functions on measurement noise variance and the optimal estimation error value of the Kalman filter 2;
(2) Measuring the real-time estimation and error compensation of the noise variance: under the condition that the measurement noise is non-Gaussian noise, the LSTM neural network in the subsystem 1 is used for predicting the variance matrix of the measurement noise in real time, the predicted measurement variance matrix is sent to the CKF in the subsystem 2, and meanwhile, the LSTM neural network in the subsystem 2 predicts and corrects the optimal estimation error value of the CKF;
(3) And (3) correcting attitude, speed and position information of the carrier/target output by the INS by using the CKF optimal estimation value after error correction in the step (2), and finally realizing that the INS/CNS integrated navigation system can output high-precision navigation information under a non-Gaussian measurement noise environment.
The process of the step (1) comprises the following steps:
(11) In the initial training stage, the subsystem 1 takes a platform misalignment angle, a speed error, a position error, a gyro constant drift and an accelerometer constant bias as state quantities, the difference between attitude, position and apparent height data solved by the CNS and attitude, position and apparent height data solved by the INS is taken as system observation measurement, optimal values of attitude error delta theta, speed error delta v and position error delta P of INS output data are filtered and estimated by using the MCGHCKF, and then a filter gain sequence [ K ] obtained at the first n moments in the filtering process is used k-1 ,K k-2 ,...,K k-n ]And a measurement noise variance matrix R obtained in the filtering process at the current moment k Respectively as input and output samples to train LSTM-1, processing and weight updating input information by a forgetting gate in an LSTM-1 neural network in the training process, and resolving by an output gateOutputting the optimal learning training result, wherein the output is R' k ；
(12) The subsystem 2 also takes a platform misalignment angle, a speed error, a position error, a gyro constant drift and an accelerometer constant offset as state quantities, and the difference between the posture, the position and the apparent height data solved by the CNS and the posture, the position and the apparent height data solved by the INS as a system observation measurement, and the optimal training result R 'of the LSTM-1 neural network is obtained' k Inputting the data into a CKF in a subsystem 2, and performing optimal filtering estimation on the system error of the INS by using the CKF to obtain optimal values of an attitude error delta theta ', a speed error delta v ' and a position error delta P ';
(13) Subtracting the attitude error, the speed error and the position error obtained by the filtering process of the subsystem 1 and the subsystem 2 to obtain delta theta ', delta v ' and delta P ':
Figure BDA0003198804950000091
then the filter gain K 'of the current moment obtained in the filtering process of the subsystem 2 is obtained' k And [ δ θ ', δ P', δ v ″ ]]The learning training samples are respectively used as input and output samples of the LSTM-2 to be trained, the input information is processed and the weight is updated through a forgetting gate in the LSTM-2 neural network in the training process, and the optimal learning training result is output after the input information is solved through an output gate.
The process of the step (2) comprises the following steps:
when the measurement noise is non-Gaussian noise, the MCGHCKF in the subsystem 1 does not work any more, and the filter gain sequence [ K 'of the first n moments obtained by the CKF in the subsystem 2' k-1 ,K′ k-2 ,...,K′ k-n ]Inputting the measured noise into the LSTM-1 after training, and predicting a variance matrix R 'of the measured noise at the current time by the LSTM-1' k (ii) a CKF in subsystem 2 is then based on R' k Filtering estimation on INS system errors delta theta ', delta v ' and delta P ' can be completed, and filtering gain K ' of the current moment is obtained ' k Prepared from K' k As the input of LSTM-2, the LSTM-2 can predict an error value delta theta' of the optimal estimation of the CKF at the current moment,δ v "and δ P"; finally, correcting INS system errors estimated by CKF by using the predicted error values to obtain corrected attitude errors delta theta 1 Velocity error δ v 1 And position error deltaP 1 。
Figure BDA0003198804950000092
The process of the step (3) comprises the following steps:
using the CKF optimal estimated value delta theta after error correction in the step (2) 1 、δv 1 And δ P 1 Attitude information θ of carrier/target solved by INS INS (k) Velocity information v INS (k) And position information P INS (k) Correction is performed so that the INS/CNS integrated navigation system outputs highly accurate attitude information θ' INS (k) And speed information v' INS (k) And position information P' INS (k)。
Figure BDA0003198804950000101
The specific process of the steps is as follows:
(1) As shown in FIG. 1, in the initial training stage, the subsystem 1 takes the platform misalignment angle, the speed error, the position error, the gyro constant drift and the accelerometer constant bias as state quantities, and the attitude θ of the CNS output CNS (k) Position P CNS (k) Apparent height h CNS (k) Information and INS output attitude θ INS (k) Position P INS (k) And a height of view h INS (k) And (4) taking the difference value of the information as an observed quantity, and establishing an INS/CNS integrated navigation system model shown in the formula (7).
Figure BDA0003198804950000102
Wherein the content of the first and second substances,
Figure BDA0003198804950000103
representing n-dimensional shapesThe vector of states is a vector of states,
Figure BDA0003198804950000104
represents an m-dimensional measurement vector, and
Figure BDA0003198804950000105
f (-) represents a nonlinear system function, w k-1 And v k Respectively represent process noise and measurement noise, and satisfy E (w) k )＝q k ，
Figure BDA0003198804950000106
E(v k )＝r k And
Figure BDA0003198804950000107
δ kj representing the kronecker function.
And optimally estimating the attitude error delta theta, the speed error delta v and the position error delta P of the INS output data by adopting an MCGHCKF algorithm according to the INS/CNS integrated navigation system equation. The volume point is calculated by Cholesky decomposition of the state covariance:
S k-1|k-1 ＝Chol(P k-1 ) (8)
Figure BDA0003198804950000111
wherein Chol (. Cndot.) represents Cholesky decomposition,
Figure BDA0003198804950000112
[1] i a set of representations [1]The ith column;
Figure BDA0003198804950000113
indicating the state optimum estimate at time k-1.
Calculating a propagation volume point through a nonlinear system function to obtain:
Figure BDA0003198804950000114
predicting the state value at the moment k to obtain:
Figure BDA0003198804950000115
wherein the content of the first and second substances,
Figure BDA0003198804950000116
the predicted value of the state covariance at time k can be expressed as:
Figure BDA0003198804950000117
generating a volume point X according to a predicted value of the state covariance at the time k i,k|k-1 ：
S k|k-1 ＝Chol(P k|k-1 ) (13)
Figure BDA0003198804950000118
The volume points are transferred by a metrology function:
Z i,k|k-1 ＝HX i,k|k-1 (15)
updating and estimating a variance matrix of the measurement noise according to a maximum correlation entropy criterion:
Figure BDA0003198804950000119
wherein M is r,k By comparing R before update k Is obtained by performing Cholesky decomposition on the raw materials,
Figure BDA0003198804950000121
G σ (. Represents a Gaussian kernel function, e) i,k ＝d i,k -g i (x k )，d i,k Is shown by D k I element of (2), g i (x k ) Denotes g (x) k ) The (i) th element of (2),
Figure BDA0003198804950000122
Figure BDA0003198804950000123
by predicting the measurement value and the measurement value covariance at the time k using equations (17) and (18), respectively, it is possible to obtain:
Figure BDA0003198804950000124
Figure BDA0003198804950000125
and predicting the cross covariance matrix at the k moment:
Figure BDA0003198804950000126
according to P zz,k|k-1 And P xz,k|k-1 Calculating a filtering gain at the k moment:
Figure BDA0003198804950000127
the state estimate at time k
Figure BDA0003198804950000128
Sum state covariance estimate P k Can be represented as:
Figure BDA0003198804950000129
Figure BDA00031988049500001210
wherein the estimated state value
Figure BDA00031988049500001211
The error amounts such as the INS attitude error δ θ, the velocity error δ v, the position error δ P, and the like are included.
In the initial training stage, the filtering process of the MCGHCKF algorithm is continuously updated and executed, and then the filtering gain sequence [ K ] obtained at the first n moments in the filtering process is obtained k-1 ,K k-2 ,...,K k-n ]And a measurement noise variance matrix R obtained in the filtering process at the current moment k The LSTM-1 is trained as an input sample and an output sample respectively, input information is processed and weight is updated through a forgetting gate in an LSTM-1 neural network in the training process, and an optimal learning training result is output after the input information is resolved through an output gate.
Simultaneously training the LSTM-1, and optimizing a training result R 'of the LSTM-1 neural network' k And (3) inputting the system error into the CKF in the subsystem 2, and performing optimal filtering estimation on the system error of the INS by using the CKF by the subsystem 2 by using a system equation shown in a formula (7). By the pair-state covariance P' k-1 The volume point was calculated by performing Cholesky decomposition:
S′ k-1 ＝Chol(P′ k-1 ) (23)
Figure BDA0003198804950000131
wherein Chol (. Cndot.) represents Cholesky decomposition,
Figure BDA0003198804950000132
[1] i a set of representations [1]The ith column;
Figure BDA0003198804950000133
indicating the state optimum estimate at time k-1.
Calculating the propagation volume point by the nonlinear state function, we can get:
Figure BDA0003198804950000134
predicting the state value at the time k to obtain:
Figure BDA0003198804950000135
wherein the content of the first and second substances,
Figure BDA0003198804950000136
the predicted value of the state covariance at time k can be expressed as:
Figure BDA0003198804950000137
generating a volume point X 'according to the state predicted value at the moment k' i,k|k-1 。
S′ k|k-1 ＝Chol(P′ k|k-1 ) (28)
Figure BDA0003198804950000141
The volume points are transferred by a metrology function:
Z′ i,k|k-1 ＝HX′ i,k|k-1 (30)
the measurement value and the measurement value covariance at the time k are respectively predicted by equations (31) and (32), and the following results are obtained:
Figure BDA0003198804950000142
Figure BDA0003198804950000143
and predicting the cross covariance matrix at the k moment:
Figure BDA0003198804950000144
according to P' zz,k|k-1 And P' xz,k|k-1 Calculating filter gain K 'at moment K' k 。
K′ k ＝P′ xz,k|k-1 (P′ zz,k|k-1 ) -1 (34)
The state estimate at time k can be expressed as:
Figure BDA0003198804950000145
the state covariance estimate at time k can be expressed as:
P′ k ＝P′ k|k-1 -K′ k P′ zz,k|k-1 (K′ k ) T (36)
the optimal estimation value of the system state vector at the moment k can be obtained according to the CKF filtering process
Figure BDA0003198804950000146
It is composed of
Figure BDA0003198804950000147
The INS error amounts include an attitude error δ θ ', a velocity error δ v ', a position error δ P ', and the like.
And (3) subtracting the attitude error, the speed error and the position error obtained through the filtering process of the subsystem 1 and the subsystem 2 to obtain delta theta ', delta v ' and delta P ':
Figure BDA0003198804950000151
then the filter gain K 'of the current moment obtained in the filtering process of the subsystem 2 is obtained' k And [ δ θ ', δ P', δ v ″ ]]Respectively as input and output samples of LSTM-2, and passing through LSTM-2 neural network during trainingAnd the forgetting gate processes the input information and updates the weight, and outputs an optimal learning training result after the resolving of the output gate.
(2) As shown in fig. 2, during the real-time estimation and error compensation steps of the variance of the measured noise. When the measurement noise is non-Gaussian noise, the MCGHCKF in the subsystem 1 does not work any more, and the filter gain sequence [ K 'of the first n time moments obtained by the CKF in the subsystem 2 is used' k-1 ,K′ k-2 ,...,K′ k-n ]Inputting the measured noise into the LSTM-1 after training, and predicting a variance matrix R 'of the measured noise at the current time by the LSTM-1' k (ii) a CKF in subsystem 2 is then R' k The filtering estimation of INS system errors delta theta ', delta v ' and delta P ' can be finished, and the filtering gain K ' of the current moment is obtained ' k Prepared from K' k As the input of LSTM-2, the error values δ θ ", δ v", and δ P "of the optimal estimation of CKF at the current time can be predicted by LSTM-2; finally, correcting INS system errors estimated by CKF by using the predicted error values to obtain corrected attitude errors delta theta 1 Velocity error δ v 1 And position error δ P 1 。
Figure BDA0003198804950000152
(3) Using the optimal estimated value delta theta of CKF after error correction in step (2) 1 、δv 1 And δ P 1 Attitude information θ of carrier/target solved by INS INS (k) Velocity information v INS (k) And position information P INS (k) Correcting to make INS/CNS integrated navigation system output high-precision attitude information theta' INS (k) And speed information v' INS (k) And position information P' INS (k)。
Figure BDA0003198804950000161
The INS/CNS integrated navigation method applicable to the non-Gaussian measurement noise environment is provided in the process, and the problem that the INS/CNS integrated navigation system is low in navigation accuracy in the non-Gaussian measurement noise environment is solved.
Compared with an INS/CNS integrated navigation algorithm based on an information theory optimization criterion, the execution efficiency of the navigation process is obviously improved.
Example two:
it is an object of this embodiment to provide an INS/CNS integrated navigation system comprising:
a prediction module configured to: acquiring attitude, speed and position information of a target, and acquiring a predicted value of non-Gaussian measurement noise variance and an optimal estimation error in the attitude, speed and position information based on a trained neural network;
a compensation module configured to: correcting the optimal estimation error value in the INS/CNS integrated navigation system based on the non-Gaussian measured noise variance and the predicted value of the optimal estimation error;
an output module configured to: and correcting the output target attitude, speed and position information based on the corrected optimal estimation error value, thereby obtaining the navigation information output by the INS/CNS integrated navigation system.
The INS/CNS integrated navigation method can be suitable for non-Gaussian measurement noise environments, and solves the problem that the INS/CNS integrated navigation system is low in navigation accuracy in the non-Gaussian measurement noise environments. Compared with an INS/CNS integrated navigation algorithm based on an information theory optimization criterion, the execution efficiency of the navigation process is obviously improved.
Example three:
the present embodiment provides a computer-readable storage medium, on which a computer program is stored, which when executed by a processor implements the steps in the INS/CNS combination navigation method according to the first embodiment.
The INS/CNS integrated navigation method can be suitable for the non-Gaussian measured noise environment, and solves the problem of low navigation accuracy of the INS/CNS integrated navigation system in the non-Gaussian measured noise environment. Compared with an INS/CNS integrated navigation algorithm based on an information theory optimization criterion, the execution efficiency of the navigation process is obviously improved.
Example four:
the present embodiment provides a computer apparatus comprising a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the combined INS/CNS navigation method according to the first embodiment.
The INS/CNS integrated navigation method can be suitable for non-Gaussian measurement noise environments, and solves the problem that the INS/CNS integrated navigation system is low in navigation accuracy in the non-Gaussian measurement noise environments. Compared with an INS/CNS integrated navigation algorithm based on an information theory optimization criterion, the execution efficiency of the navigation process is obviously improved.
The steps or modules related to the second to fourth embodiments correspond to those of the first embodiment, and the detailed description thereof can be found in the relevant description section of the first embodiment. The term "computer-readable storage medium" should be taken to include a single medium or multiple media containing one or more sets of instructions; it should also be understood to include any medium that is capable of storing, encoding or carrying a set of instructions for execution by a processor and that cause the processor to perform any of the methods of the present invention.
The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention, and various modifications and changes may be made by those skilled in the art. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims (4)

  1. An INS/CNS integrated navigation method, characterized by: the method comprises the following steps:
    acquiring attitude, speed and position information of a target, and acquiring a predicted value of non-Gaussian measurement noise variance and an optimal estimation error in the attitude, speed and position information based on a trained neural network;
    the training process of the neural network comprises the steps that a first subsystem of the INS/CNS integrated navigation system takes a platform misalignment angle, a speed error, a position error, gyro constant drift and accelerometer constant bias as state quantities, and the difference between attitude, position and apparent height data output by the astronomical navigation system and attitude, position and apparent height data output by the inertial navigation system is taken as an observed quantity, so that an INS/CNS integrated navigation system model is constructed;
    based on an INS/CNS integrated navigation system model, acquiring optimal filtering estimation values of attitude errors, speed errors and position errors in output data of an inertial navigation system by using a Kalman filtering algorithm of maximum correlation entropy; respectively taking a filter gain sequence obtained at the first n moments in the filtering process and a measurement noise variance matrix obtained in the filtering process at the current moment as input and output samples to train a neural network of the first subsystem, and outputting an optimal learning training result, wherein n is more than or equal to 1;
    the second subsystem of the INS/CNS integrated navigation system takes a platform misalignment angle, a speed error, a position error, a gyro constant drift and an accelerometer constant bias as state quantities, the difference between attitude, position and apparent height data output by an astronomical navigation system and attitude, position and apparent height data output by an inertial navigation system is taken as system view measurement, the optimal training result of a neural network of the first subsystem is input into a volume Kalman filter of the second subsystem, and the optimal filtering estimation is carried out on the error of the inertial navigation system to obtain the optimal values of the attitude error, the speed error and the position error;
    obtaining the difference values of attitude errors, speed errors and position errors of the first subsystem and the second subsystem, respectively serving as input and output samples of a neural network of the second subsystem for training based on the difference values and the filtering gain of the second subsystem at the current moment, and outputting an optimal learning training result;
    correcting the optimal estimation error value in the INS/CNS integrated navigation system based on the non-Gaussian measured noise variance and the predicted value of the optimal estimation error; under the condition that the measurement noise is non-Gaussian noise, the filter gain sequence of the first n moments in the second subsystem is input into the trained neural network of the first subsystem, and a variance matrix of the measurement noise at the current moment is predicted, wherein n is more than or equal to 1; the second subsystem obtains a filtering estimation value of a system error based on a variance matrix of the measured noise at the current moment, obtains a filtering gain at the current moment, takes the filtering gain as the input of a neural network of the second subsystem, and predicts an error value of optimal estimation at the current moment; correcting the system error of the inertial navigation system based on the predicted error value to obtain a corrected attitude error, a corrected speed error and a corrected position error;
    and correcting the output target attitude, speed and position information based on the corrected optimal estimated error value, thereby obtaining the navigation information output by the INS/CNS integrated navigation system.
  2. An ins/CNS integrated navigation system comprising:
    a prediction module configured to: acquiring attitude, speed and position information of a target, and acquiring a non-Gaussian measurement noise variance and a predicted value of an optimal estimation error in the attitude, speed and position information based on a trained neural network;
    the training process of the neural network comprises the steps that a first subsystem of the INS/CNS integrated navigation system takes a platform misalignment angle, a speed error, a position error, gyro constant drift and accelerometer constant bias as state quantities, and the difference between attitude, position and apparent height data output by the astronomical navigation system and attitude, position and apparent height data output by the inertial navigation system is observed quantities to construct an INS/CNS integrated navigation system model;
    based on an INS/CNS integrated navigation system model, obtaining optimal filtering estimation values of attitude errors, speed errors and position errors in output data of an inertial navigation system by using a Kalman filtering algorithm of the maximum correlation entropy; respectively taking a filter gain sequence obtained at the first n moments in the filtering process and a measurement noise variance matrix obtained in the filtering process at the current moment as input and output samples to train a neural network of the first subsystem, and outputting an optimal learning training result, wherein n is more than or equal to 1;
    the method comprises the following steps that a second subsystem of the INS/CNS integrated navigation system takes a platform misalignment angle, a speed error, a position error, a gyro constant drift and an accelerometer constant bias as state quantities, the difference between attitude, position and apparent height data output by an astronomical navigation system and attitude, position and apparent height data output by an inertial navigation system is taken as system view measurement, the optimal training result of a neural network of the first subsystem is input into a volume Kalman filter of the second subsystem, the error of the inertial navigation system is subjected to optimal filtering estimation, and the optimal values of the attitude error, the speed error and the position error are obtained;
    acquiring difference values of attitude errors, speed errors and position errors of the first subsystem and the second subsystem, respectively taking the difference values and the filter gains of the second subsystem at the current moment as input and output samples of a neural network of the second subsystem for training, and outputting an optimal learning training result;
    a compensation module configured to: correcting the optimal estimation error value in the INS/CNS integrated navigation system based on the non-Gaussian measured noise variance and the predicted value of the optimal estimation error; under the condition that the measurement noise is non-Gaussian noise, the filter gain sequence of the first n moments in the second subsystem is input into the trained neural network of the first subsystem, and a variance matrix of the measurement noise at the current moment is predicted, wherein n is more than or equal to 1; the second subsystem obtains a filtering estimation value of a system error based on a variance matrix of the measured noise at the current moment, obtains a filtering gain at the current moment, takes the filtering gain as the input of a neural network of the second subsystem, and predicts an error value of optimal estimation at the current moment; correcting the system error of the inertial navigation system based on the predicted error value to obtain a corrected attitude error, a corrected speed error and a corrected position error;
    an output module configured to: and correcting the output target attitude, speed and position information based on the corrected optimal estimated error value, thereby obtaining the navigation information output by the INS/CNS integrated navigation system.
  3. 3. A computer readable storage medium having stored thereon a computer program which, when executed by a processor, performs the steps in the INS/CNS combination navigation method according to claim 1.
  4. 4. A computer device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, wherein the processor when executing the program implements the steps in the INS/CNS combination navigation method according to claim 1.
CN202110898503.5A 2021-08-05 2021-08-05 INS/CNS integrated navigation method, system, storage medium and equipment Active CN113447019B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110898503.5A CN113447019B (en) 2021-08-05 2021-08-05 INS/CNS integrated navigation method, system, storage medium and equipment

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110898503.5A CN113447019B (en) 2021-08-05 2021-08-05 INS/CNS integrated navigation method, system, storage medium and equipment

Publications (2)

Publication Number Publication Date
CN113447019A CN113447019A (en) 2021-09-28
CN113447019B true CN113447019B (en) 2023-01-13

Family

ID=77818215

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110898503.5A Active CN113447019B (en) 2021-08-05 2021-08-05 INS/CNS integrated navigation method, system, storage medium and equipment

Country Status (1)

Country Link
CN (1) CN113447019B (en)

Family Cites Families (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101826852A (en) * 2010-03-11 2010-09-08 哈尔滨工程大学 Unscented particle filtering method based on particle swarm optimization algorithm
CN103363993B (en) * 2013-07-06 2016-04-20 西北工业大学 A kind of aircraft angle rate signal reconstructing method based on Unscented kalman filtering
CN105424030B (en) * 2015-11-24 2018-11-09 东南大学 Fusion navigation device and method based on wireless fingerprint and MEMS sensor
CN105737832B (en) * 2016-03-22 2019-03-22 北京工业大学 Distributed SLAM method based on global optimum's data fusion
CN107390246A (en) * 2017-07-06 2017-11-24 电子科技大学 A kind of GPS/INS Combinated navigation methods based on genetic neural network
CN108562290B (en) * 2018-07-13 2020-10-27 深圳市戴升智能科技有限公司 Navigation data filtering method and device, computer equipment and storage medium
CN109459040B (en) * 2019-01-14 2021-06-18 哈尔滨工程大学 Multi-AUV (autonomous Underwater vehicle) cooperative positioning method based on RBF (radial basis function) neural network assisted volume Kalman filtering
CN110222783B (en) * 2019-06-13 2023-04-07 南京信息工程大学 Foundation and satellite-borne radar precipitation data fusion method based on wavelet domain regularization
CN110632634B (en) * 2019-09-24 2022-11-01 东南大学 Optimal data fusion method suitable for ballistic missile INS/CNS/GNSS combined navigation system
CN111156987B (en) * 2019-12-18 2022-06-28 东南大学 Inertia/astronomy combined navigation method based on residual compensation multi-rate CKF

Also Published As

Publication number Publication date
CN113447019A (en) 2021-09-28

Similar Documents

Publication Publication Date Title
CN111156987B (en) Inertia/astronomy combined navigation method based on residual compensation multi-rate CKF
CN108827310B (en) Marine star sensor auxiliary gyroscope online calibration method
CN109974714B (en) Sage-Husa self-adaptive unscented Kalman filtering attitude data fusion method
CN112146655B (en) Elastic model design method for BeiDou/SINS tight integrated navigation system
Shin et al. Adaptive support vector regression for UAV flight control
CN112945225A (en) Attitude calculation system and method based on extended Kalman filtering
CN111141313B (en) Method for improving matching transfer alignment precision of airborne local relative attitude
CN112798021B (en) Inertial navigation system inter-travelling initial alignment method based on laser Doppler velocimeter
CN113155129B (en) Holder attitude estimation method based on extended Kalman filtering
CN110044385B (en) Rapid transfer alignment method under condition of large misalignment angle
CN110595434B (en) Quaternion fusion attitude estimation method based on MEMS sensor
Wang et al. Improved Kalman filter and its application in initial alignment
Sun et al. A robust indirect Kalman filter based on the gradient descent algorithm for attitude estimation during dynamic conditions
CN113587926A (en) Spacecraft space autonomous rendezvous and docking relative navigation method
CN113447019B (en) INS/CNS integrated navigation method, system, storage medium and equipment
Sang et al. Invariant cubature kalman filtering-based visual-inertial odometry for Robot pose estimation
CN111854741A (en) GNSS/INS tight combination filter and navigation method
CN110912535A (en) Novel pilot-free Kalman filtering method
CN113447018B (en) Real-time attitude estimation method of underwater inertial navigation system
Chen et al. Dealing with observation outages within navigation data using Gaussian process regression
CN114323007A (en) Carrier motion state estimation method and device
Cao et al. Adaptive predictive variable structure filter for attitude synchronization estimation
Girrbach et al. Adaptive compensation of measurement delays in multi-sensor fusion for inertial motion tracking using moving horizon estimation
CN110543919A (en) robot positioning control method, terminal equipment and storage medium
ZHOU et al. Novel SCNS/RSINS tight-integrated alignment based on adaptive interacting multiple model filters on shipboard

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
GR01 Patent grant
GR01 Patent grant