CN116538872A - Inertial navigation trajectory parameter measurement method based on Conv-LSTM and Kalman filtering fusion - Google Patents
Inertial navigation trajectory parameter measurement method based on Conv-LSTM and Kalman filtering fusion Download PDFInfo
- Publication number
- CN116538872A CN116538872A CN202310692166.3A CN202310692166A CN116538872A CN 116538872 A CN116538872 A CN 116538872A CN 202310692166 A CN202310692166 A CN 202310692166A CN 116538872 A CN116538872 A CN 116538872A
- Authority
- CN
- China
- Prior art keywords
- trajectory
- ballistic
- conv
- lstm
- kalman filtering
- 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
- 238000001914 filtration Methods 0.000 title claims abstract description 38
- 230000004927 fusion Effects 0.000 title claims abstract description 17
- 238000000691 measurement method Methods 0.000 title claims abstract description 16
- 238000000034 method Methods 0.000 claims abstract description 25
- 238000005259 measurement Methods 0.000 claims abstract description 24
- 238000012549 training Methods 0.000 claims description 20
- 238000010606 normalization Methods 0.000 claims description 18
- 238000012360 testing method Methods 0.000 claims description 15
- 230000006870 function Effects 0.000 claims description 9
- 238000012545 processing Methods 0.000 claims description 8
- 230000004913 activation Effects 0.000 claims description 6
- 230000003068 static effect Effects 0.000 claims description 6
- 230000009977 dual effect Effects 0.000 claims description 5
- 238000004364 calculation method Methods 0.000 claims description 4
- 230000001133 acceleration Effects 0.000 claims description 3
- 210000002569 neuron Anatomy 0.000 claims description 3
- 230000036544 posture Effects 0.000 description 12
- RZVHIXYEVGDQDX-UHFFFAOYSA-N 9,10-anthraquinone Chemical compound C1=CC=C2C(=O)C3=CC=CC=C3C(=O)C2=C1 RZVHIXYEVGDQDX-UHFFFAOYSA-N 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000005422 blasting Methods 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000005358 geomagnetic field Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000007500 overflow downdraw method Methods 0.000 description 1
- 238000012876 topography Methods 0.000 description 1
- 230000035899 viability Effects 0.000 description 1
Classifications
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F42—AMMUNITION; BLASTING
- F42B—EXPLOSIVE CHARGES, e.g. FOR BLASTING, FIREWORKS, AMMUNITION
- F42B35/00—Testing or checking of ammunition
-
- 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
- G01C21/10—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
- G01C21/183—Compensation of inertial measurements, e.g. for temperature effects
- G01C21/188—Compensation of inertial measurements, e.g. for temperature effects for accumulated errors, e.g. by coupling inertial systems with absolute positioning systems
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S19/00—Satellite radio beacon positioning systems; Determining position, velocity or attitude using signals transmitted by such systems
- G01S19/38—Determining a navigation solution using signals transmitted by a satellite radio beacon positioning system
- G01S19/39—Determining a navigation solution using signals transmitted by a satellite radio beacon positioning system the satellite radio beacon positioning system transmitting time-stamped messages, e.g. GPS [Global Positioning System], GLONASS [Global Orbiting Navigation Satellite System] or GALILEO
- G01S19/42—Determining position
- G01S19/48—Determining position by combining or switching between position solutions derived from the satellite radio beacon positioning system and position solutions derived from a further system
- G01S19/49—Determining position by combining or switching between position solutions derived from the satellite radio beacon positioning system and position solutions derived from a further system whereby the further system is an inertial position system, e.g. loosely-coupled
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/044—Recurrent networks, e.g. Hopfield networks
- G06N3/0442—Recurrent networks, e.g. Hopfield networks characterised by memory or gating, e.g. long short-term memory [LSTM] or gated recurrent units [GRU]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/0464—Convolutional networks [CNN, ConvNet]
-
- H—ELECTRICITY
- H03—ELECTRONIC CIRCUITRY
- H03H—IMPEDANCE NETWORKS, e.g. RESONANT CIRCUITS; RESONATORS
- H03H17/00—Networks using digital techniques
- H03H17/02—Frequency selective networks
- H03H17/0248—Filters characterised by a particular frequency response or filtering method
- H03H17/0255—Filters based on statistics
- H03H17/0257—KALMAN filters
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T90/00—Enabling technologies or technologies with a potential or indirect contribution to GHG emissions mitigation
Landscapes
- Engineering & Computer Science (AREA)
- Remote Sensing (AREA)
- Radar, Positioning & Navigation (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Mathematical Physics (AREA)
- Evolutionary Computation (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Automation & Control Theory (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Probability & Statistics with Applications (AREA)
- Computer Hardware Design (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Biology (AREA)
- Computer Networks & Wireless Communication (AREA)
- Navigation (AREA)
Abstract
The invention relates to an inertial navigation trajectory parameter measurement method based on Conv-LSTM and Kalman filtering fusion, and belongs to the technical field of ammunition inertial measurement. The method can realize high-precision real-time ballistic measurement of the shot intelligent ammunition under the condition of satellite positioning failure caused by complex electromagnetic environment or enemy interference and deception. The method is mainly aimed at an intelligent ammunition which carries out ballistic measurement by means of inertia/satellite combined signals, can solve the problems that measurement errors are accumulated along with time and the like when the satellite positioning is invalid in the traditional integrated navigation algorithm, and has higher precision.
Description
Technical Field
The invention relates to an inertial navigation trajectory parameter measurement method based on Conv-LSTM and Kalman filtering fusion, and belongs to the technical field of ammunition inertial measurement.
Background
Accurate ballistic measurements are an important prerequisite for intelligent ammunition flight control and accurate striking for shot blasting. Satellite positioning systems are currently the dominant method of intelligent ammunition ballistic measurement, however, satellite signals are susceptible to electromagnetic interference and the availability under combat conditions is severely reduced. The development of a ballistic measurement method under satellite refusal conditions has important significance for improving the viability of intelligent ammunition. Triaxial geomagnetic sensors can be used for measuring the attitude of an projectile, but are extremely susceptible to strong magnetic environment interference due to weak geomagnetic field signals. The wireless telecommunication beacon technology is widely applied to aircraft navigation in the civil field, however, in the actual battlefield environment, the number, the range and the topography of the beacon layout are limited, and often, the complete three-dimensional measurement information with long time and high precision cannot be provided, so that the observability of the estimated projectile attitude is poor. In recent years, the application of the micro-electromechanical inertial sensor with low cost, small volume and strong overload resistance to intelligent ammunition is gradually possible, however, the error of the low-cost inertial navigation system is accumulated with time, and the error is often dependent on other auxiliary navigation systems.
Disclosure of Invention
The technical solution of the invention is as follows: the method is an artificial intelligence based projectile track estimation method which only uses micro-electromechanical inertial sensors and ammunition specific pre-flight parameters in a satellite rejection environment, aims to solve the problem that measurement errors are accumulated over time in a traditional inertial navigation system in the satellite rejection environment, and can improve the accuracy of ballistic measurement in the satellite rejection environment.
The technical scheme of the invention is as follows:
an inertial navigation trajectory parameter measurement method based on Conv-LSTM and Kalman filtering fusion comprises the following steps:
firstly, building a training data set, wherein the built training data set comprises ballistic basic parameters and analog noise output data of an inertial sensor;
the trajectory basic parameters include a trajectory angle a, a trajectory initial position obtained under the condition of simulating the trajectory angle a, a trajectory initial speed, a trajectory initial posture, a trajectory position at each moment, a trajectory speed at each moment, and a trajectory posture at each moment;
the acquisition method of the analog noise output data of the inertial sensor comprises the following steps: inverting the trajectory basic parameters to obtain noiseless output data of the inertial sensor, and introducing noise uniformly distributed in a set range into the noiseless output data of the inertial sensor to obtain analog noise output data of the inertial sensor;
the setting range is as follows: [ -f r ,f r ]、[-ω r ,ω r ]，f r Is zero bias stability index omega of accelerometer in inertial sensor r The zero offset stability index of the gyroscope in the inertial sensor;
step two, a test data set is established, wherein the established test data set comprises trajectory basic parameters and actual measurement noise output data of the inertial sensor;
the trajectory basic parameters include a trajectory angle B, a trajectory initial position, a trajectory initial velocity, a trajectory initial posture, a trajectory position at each moment, a trajectory velocity at each moment, and a trajectory posture at each moment obtained under the condition of simulating the trajectory angle B; the minimum value of the ballistic angle B value range is smaller than the minimum value of the ballistic angle A value range, and the maximum value of the ballistic angle B value range is larger than the maximum value of the ballistic angle A value range;
the acquisition method of the output data of the measured noise of the inertial sensor comprises the following steps: inverting the trajectory basic parameters to obtain noiseless output data of the inertial sensor, and then introducing actual measurement static noise data into the noiseless output data of the inertial sensor to obtain actual measurement noise output data of the inertial sensor;
thirdly, normalizing the training data set established in the first step by using a min-max normalization method to obtain a normalized training data set, and normalizing the test data set established in the second step by using the min-max normalization method to obtain a normalized test data set;
step four, a Conv-LSTM network is established, the established Conv-LSTM network comprises a one-dimensional residual error network, a double LSTM network and a parameter regression layer, wherein the one-dimensional residual error network comprises eighteen basic modules, and each basic module comprises 2 convolution units with convolution kernels of 3, a Relu activation function, batch normalization, a dropout layer and a residual error calculation module; the one-dimensional residual error network is used for learning hidden variables of the ballistic equation;
the dual LSTM network comprises two standard LSTM networks, a ReLU activation function and a dropout layer; each standard LSTM network comprises 5-10 neurons, and the double LSTM network is used for fusing the current hidden state and the previous hidden state so as to estimate the optimal current hidden motion state;
the parameter regression layer comprises three branches, namely a branch I, a branch II and a branch III, wherein each branch is divided into two small branches, one small branch in the branch I is used for regressing the position information of the current moment, the other small branch in the branch I is used for regressing the position covariance information of the current moment, one small branch in the branch II is used for regressing the speed information of the current moment, the other small branch in the branch II is used for regressing the speed covariance information of the current moment, one small branch in the branch III is used for regressing the attitude information of the current moment, and the other small branch in the branch I is used for regressing the attitude covariance information of the current moment;
the built Conv-LSTM network loss function is designed to minimize the mean square error of the predicted and actual quantities;
the built optimizer of Conv-LSTM network is SGD optimizing model;
fifthly, training the Conv-LSTM network established in the fourth step by using the training data set obtained in the third step after normalization processing to obtain a trained Conv-LSTM network;
step six, storing the trained Conv-LSTM network obtained in the step five on a missile-borne computer, and acquiring and storing real-time output data measured by an inertial sensor after the missile emits;
a seventh step of inputting the real-time output data of the inertial sensor obtained in the sixth step or the normalized test data set obtained in the third step into the Conv-LSTM network trained in the fifth step to obtain predicted trajectory pose parameters, wherein the trajectory pose parameters comprise trajectory positions, trajectory speeds, trajectory postures, trajectory position covariance values, trajectory speed covariance values and trajectory posture covariance values;
eighth step, an extended Kalman filtering model is established;
the observed quantity of the established extended Kalman filtering model is the predicted ballistic position, ballistic speed and ballistic attitude obtained in the seventh step;
the observed variance value of the established extended Kalman filtering model is the predicted ballistic position variance value, the predicted ballistic velocity variance value and the predicted ballistic attitude variance value obtained in the seventh step;
the state quantity of the established extended Kalman filtering model is the ballistic position, the ballistic speed, the ballistic posture, the ballistic acceleration and the ballistic angular speed at the current moment;
the established state equation of the extended Kalman filtering model is an inertial navigation solution model;
and a ninth step, taking the predicted ballistic position, ballistic speed and ballistic gesture obtained in the seventh step as the initial state value of the extended Kalman filtering model obtained in the eighth step, and updating the state of the extended Kalman filtering model to obtain optimized ballistic position, ballistic speed and ballistic gesture data.
The invention discloses a pure inertial navigation trajectory parameter measurement method based on Conv-LSTM and Kalman filtering fusion. Compared with the prior art, the method has the following advantages and beneficial effects:
1. the method for measuring the ballistic parameters of the shot intelligent ammunition by using the Conv-LSTM and Kalman filtering fusion method is realized for the first time, and a physical model built based on intelligent ammunition dynamics is reduced.
2. According to the invention, ballistic prediction based on the low-cost micro inertial sensor is realized for the first time, and the cost, power consumption and weight improvement brought by adding the sensor are reduced.
3. According to the method, the Conv-LSTM network is established, the strapdown inertial navigation solution pose can be effectively predicted according to the inertial sensor measurement data and the ballistic parameters before the launching, the accumulated errors are corrected, and the ballistic measurement accuracy is further improved.
4. According to the invention, a Kalman filtering model is established, conv-LSTM network output is used as an observed quantity, so that the measurement precision of a pure inertial navigation system is further improved, further, trajectory pose parameters can be accurately obtained, trajectory drop points are accurately predicted in advance, further, a control instruction is calculated, the projectile body accurately hits a target, and the intelligent ammunition hit precision and damage effect are improved.
5. According to the invention, under the condition of satellite positioning failure caused by complex electromagnetic environment or enemy interference and deception, high-precision real-time ballistic measurement of shot intelligent ammunition can be realized, and the subsequent flight control and hit precision are improved.
Drawings
FIG. 1 is a schematic flow chart of the method of the present invention.
Detailed Description
The following describes specific embodiments of the present invention with reference to the drawings.
Examples
As shown in fig. 1, a method for measuring inertial navigation trajectory parameters based on Conv-LSTM and kalman filter fusion comprises the following steps:
firstly, building a training data set, wherein the built training data set comprises ballistic basic parameters and analog noise output data of an inertial sensor;
the ballistic basic parameters include a ballistic angle [35 °,50 ° ] (interval 1 °) and a ballistic initial position, a ballistic initial velocity, a ballistic initial posture, a ballistic position at each moment, a ballistic velocity at each moment, and a ballistic posture at each moment obtained under the condition of simulating the ballistic angle;
the acquisition method of the analog noise output data of the inertial sensor comprises the following steps: inverting the trajectory basic parameters to obtain noiseless output data of the inertial sensor, and introducing noise uniformly distributed in a set range into the noiseless output data of the inertial sensor to obtain analog noise output data of the inertial sensor;
the setting range is as follows: [ -2mg,2mg]、[-10°/h,10°/h]，f r Is zero bias stability index omega of accelerometer in inertial sensor r The zero offset stability index of the gyroscope in the inertial sensor;
step two, a test data set is established, wherein the established test data set comprises trajectory basic parameters and actual measurement noise output data of the inertial sensor;
the ballistic basic parameters include a ballistic angle [30 °,55 ° ] (interval 1 °) and a ballistic initial position, a ballistic initial velocity, a ballistic initial posture, a ballistic position at each moment, a ballistic velocity at each moment, and a ballistic posture at each moment obtained under the condition of simulating the ballistic angle;
the acquisition method of the output data of the measured noise of the inertial sensor comprises the following steps: inverting the trajectory basic parameters to obtain noiseless output data of the inertial sensor, and then introducing actual measurement static noise data into the noiseless output data of the inertial sensor to obtain actual measurement noise output data of the inertial sensor;
in the actual measurement of static noise, the zero offset of the actual measurement static noise of the 3-axis accelerometer in the inertial sensor is 1.44mg, 0.09mg and 0.99mg, and the standard deviation is 2.3388mg, 2.8898mg and 2.0049mg; the zero offset of the measured static noise of the 3-axis gyroscope in the inertial sensor is 30.00 degrees/h, 43.20 degrees/h and 2.31 degrees/h, and the standard deviation is 262.50 degrees/h, 286.1771 degrees/h and 221.4064 degrees/h;
under the data set, when the flight time is 100s, the traditional strapdown inertial calculation precision is as follows: the pitch angle, roll angle and yaw angle prediction errors are: 5.82 °, 5.86 ° and 5.23 °; the prediction errors of the east speed, the north speed and the sky speed are respectively as follows: 49.21m/s, 14.03m/s and 33.72m/s; the speed prediction errors of the east position, the north position and the sky position are respectively as follows: 2852.5m, 5096.5m and 1901.9m.
Thirdly, normalizing the training data set established in the first step by using a min-max normalization method to obtain a normalized training data set, and normalizing the test data set established in the second step by using the min-max normalization method to obtain a normalized test data set;
step four, a Conv-LSTM network is established, the established Conv-LSTM network comprises a one-dimensional residual error network, a double LSTM network and a parameter regression layer, wherein the one-dimensional residual error network comprises eighteen basic modules, and each basic module comprises 2 convolution units with convolution kernels of 3, a Relu activation function, batch normalization, a dropout layer and a residual error calculation module; the one-dimensional residual error network is used for learning hidden variables of the ballistic equation, and in the example, the one-dimensional residual error network adopts a Resnet18 model;
the dual LSTM network comprises two standard LSTM networks, a ReLU activation function and a dropout layer; each standard LSTM network comprises 5 neurons, and the dual LSTM network is used for fusing the current hidden state and the previous hidden state so as to estimate the optimal current hidden motion state;
the parameter regression layer comprises three branches, namely a branch I, a branch II and a branch III, wherein each branch is divided into two small branches, one small branch in the branch I is used for regressing the position information of the current moment, the other small branch in the branch I is used for regressing the position covariance information of the current moment, one small branch in the branch II is used for regressing the speed information of the current moment, the other small branch in the branch II is used for regressing the speed covariance information of the current moment, one small branch in the branch III is used for regressing the attitude information of the current moment, and the other small branch in the branch I is used for regressing the attitude covariance information of the current moment;
the built Conv-LSTM network loss function is designed to minimize the mean square error of the predicted and actual quantities;
the built optimizer of Conv-LSTM network is SGD optimizing model;
fifthly, training the Conv-LSTM network established in the fourth step by using the training data set obtained in the third step after normalization processing to obtain a trained Conv-LSTM network;
step six, storing the trained Conv-LSTM network obtained in the step five on a missile-borne computer, acquiring and storing real-time output data measured by an inertial sensor in real time after the missile is launched, wherein the test data set obtained in the step three after normalization is used as the real-time output data measured by the inertial sensor in the embodiment;
a seventh step of inputting the test data set obtained in the third step after normalization processing into the Conv-LSTM network after training in the fifth step to obtain predicted trajectory pose parameters, wherein the trajectory pose parameters comprise trajectory positions, trajectory speeds, trajectory poses, trajectory position covariance values, trajectory speed covariance values and trajectory pose covariance values; after Conv-Lstm network model, the prediction errors of pitch angle, roll angle and yaw angle are as follows: 5.10 °, 0.59 °, and 0.29 °; the prediction errors of the east speed, the north speed and the sky speed are respectively as follows: 2.02m/s, 1.56m/s and 1.74m/s; the speed prediction errors of the east position, the north position and the sky position are respectively as follows: 400.41m, 68.11m and 108.45m.
Eighth step, an extended Kalman filtering model is established;
the observed quantity of the established extended Kalman filtering model is the predicted ballistic position, ballistic speed and ballistic attitude obtained in the seventh step;
the observed variance value of the established extended Kalman filtering model is the predicted ballistic position variance value, the predicted ballistic velocity variance value and the predicted ballistic attitude variance value obtained in the seventh step;
the state quantity of the established extended Kalman filtering model is the ballistic position, the ballistic speed, the ballistic posture, the ballistic acceleration and the ballistic angular speed at the current moment;
the established state equation of the extended Kalman filtering model is an inertial navigation solution model;
and a ninth step, taking the predicted ballistic position, ballistic speed and ballistic gesture obtained in the seventh step as the initial state value of the extended Kalman filtering model obtained in the eighth step, and updating the state of the extended Kalman filtering model to obtain optimized ballistic position, ballistic speed and ballistic gesture data. After Kalman filtering, when the flight time is 100s, the prediction precision of the pitch angle, the roll angle and the yaw angle is as follows: 0.001 °, 0.005 ° and 0.02; the prediction errors of the east speed, the north speed and the sky speed are respectively as follows: 0.70m/s, 0.55m/s and 0.07m/s; the speed prediction errors of the east position, the north position and the sky position are respectively as follows: 131.34m, 28.49 and 59.47m.
It should be understood that the foregoing is only illustrative of the preferred embodiments of the present invention and is not intended to limit the scope of the present invention. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (10)
1. The inertial navigation trajectory parameter measurement method based on Conv-LSTM and Kalman filtering fusion is characterized by comprising the following steps of:
firstly, building a training data set;
step two, a test data set is established;
thirdly, carrying out normalization processing on the training data set established in the first step to obtain a training data set after normalization processing, and carrying out normalization processing on the test data set established in the second step to obtain a test data set after normalization processing;
fourthly, establishing a Conv-LSTM network;
fifthly, training the Conv-LSTM network established in the fourth step by using the training data set obtained in the third step after normalization processing to obtain a trained Conv-LSTM network;
step six, storing the trained Conv-LSTM network obtained in the step five on a missile-borne computer, and acquiring and storing real-time output data measured by an inertial sensor after the missile emits;
a seventh step of inputting the real-time output data of the inertial sensor obtained in real time in the sixth step into the Conv-LSTM network after training in the fifth step to obtain predicted trajectory pose parameters;
eighth step, an extended Kalman filtering model is established;
and ninth, taking the trajectory pose parameter obtained in the seventh step as a state initial value of the extended Kalman filtering model obtained in the eighth step, and updating the state of the extended Kalman filtering model to obtain optimized trajectory position, trajectory speed and trajectory pose data.
2. The inertial navigation trajectory parameter measurement method based on Conv-LSTM and Kalman filtering fusion according to claim 1, wherein the method comprises the following steps:
in the first step, the training data set established comprises ballistic basic parameters and analog noise output data of an inertial sensor;
the trajectory basic parameters include a trajectory angle a, a trajectory initial position obtained under the condition of simulating the trajectory angle a, a trajectory initial speed, a trajectory initial posture, a trajectory position at each moment, a trajectory speed at each moment, and a trajectory posture at each moment;
the acquisition method of the analog noise output data of the inertial sensor comprises the following steps: inverting the trajectory basic parameters to obtain noiseless output data of the inertial sensor, and introducing noise uniformly distributed in a set range into the noiseless output data of the inertial sensor to obtain analog noise output data of the inertial sensor;
the setting range is as follows: [ -f r ,f r ]、[-ω r ,ω r ]，f r Is zero bias stability index omega of accelerometer in inertial sensor r Is an index of zero bias stability of a gyroscope in the inertial sensor.
3. The inertial navigation trajectory parameter measurement method based on Conv-LSTM and Kalman filtering fusion according to claim 2, wherein the method is characterized by comprising the following steps:
in the second step, the established test data set comprises ballistic basic parameters and measured noise output data of the inertial sensor;
the trajectory basic parameters include a trajectory angle B, a trajectory initial position, a trajectory initial velocity, a trajectory initial posture, a trajectory position at each moment, a trajectory velocity at each moment, and a trajectory posture at each moment obtained under the condition of simulating the trajectory angle B; the minimum value of the ballistic angle B value range is smaller than the minimum value of the ballistic angle A value range, and the maximum value of the ballistic angle B value range is larger than the maximum value of the ballistic angle A value range;
the acquisition method of the output data of the measured noise of the inertial sensor comprises the following steps: inverting the trajectory basic parameters to obtain noiseless output data of the inertial sensor, and then introducing actual measurement static noise data into the noiseless output data of the inertial sensor to obtain actual measurement noise output data of the inertial sensor.
4. A method for measuring inertial navigation trajectory parameters based on Conv-LSTM and kalman filter fusion according to any one of claims 1-3, characterized by:
and in the third step, carrying out normalization treatment by using a min-max normalization method.
5. The inertial navigation trajectory parameter measurement method based on Conv-LSTM and Kalman filtering fusion according to claim 4, wherein the method comprises the following steps:
in the fourth step, the Conv-LSTM network is established to comprise a one-dimensional residual network, a double LSTM network and a parameter regression layer.
6. The inertial navigation trajectory parameter measurement method based on Conv-LSTM and Kalman filtering fusion according to claim 5, wherein the method comprises the following steps of:
the one-dimensional residual error network comprises eighteen basic modules, wherein each basic module comprises 2 convolution units with convolution kernels of 3, a Relu activation function, a batch normalization layer, a dropout layer and a residual error calculation module; the one-dimensional residual error network is used for learning hidden variables of the ballistic equation;
the dual LSTM network comprises two standard LSTM networks, a ReLU activation function and a dropout layer; each standard LSTM network includes 5-10 neurons, and the dual LSTM network is configured to fuse the current hidden state with the previous hidden state and estimate the best current hidden motion state.
7. The inertial navigation trajectory parameter measurement method based on Conv-LSTM and Kalman filtering fusion according to claim 5 or 6, wherein the method comprises the following steps:
the parameter regression layer comprises three branches, namely a branch I, a branch II and a branch III, wherein each branch is divided into two small branches, one small branch in the branch I is used for regressing the position information of the current moment, the other small branch in the branch I is used for regressing the position covariance information of the current moment, one small branch in the branch II is used for regressing the speed information of the current moment, the other small branch in the branch II is used for regressing the speed covariance information of the current moment, one small branch in the branch III is used for regressing the attitude information of the current moment, and the other small branch in the branch I is used for regressing the attitude covariance information of the current moment;
the built Conv-LSTM network loss function is designed to minimize the mean square error of the predicted and actual quantities;
the built optimizer of Conv-LSTM network is SGD optimizing model.
8. The inertial navigation trajectory parameter measurement method based on Conv-LSTM and Kalman filtering fusion of claim 7, wherein the method comprises the following steps of:
the seventh step, the trajectory pose parameter includes a trajectory position, a trajectory velocity, a trajectory pose, a trajectory position covariance value, a trajectory velocity covariance value, and a trajectory pose covariance value.
9. The inertial navigation trajectory parameter measurement method based on Conv-LSTM and Kalman filtering fusion of claim 8, wherein the method comprises the following steps of:
in the eighth step, the observed quantity of the established extended kalman filter model is the predicted ballistic position, ballistic speed and ballistic gesture;
the observed variance values of the established extended Kalman filtering model are predicted trajectory position covariance values, trajectory speed covariance values and trajectory posture covariance values;
the state quantity of the established extended Kalman filtering model is the ballistic position, the ballistic speed, the ballistic posture, the ballistic acceleration and the ballistic angular speed at the current moment;
the state equation of the established extended Kalman filtering model is an inertial navigation solution model.
10. The inertial navigation trajectory parameter measurement method based on Conv-LSTM and Kalman filtering fusion according to claim 9, wherein the method comprises the following steps:
in the ninth step, the predicted ballistic position, ballistic velocity, and ballistic pose are used as initial state values of the extended kalman filter model.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310692166.3A CN116538872A (en) | 2023-06-12 | 2023-06-12 | Inertial navigation trajectory parameter measurement method based on Conv-LSTM and Kalman filtering fusion |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310692166.3A CN116538872A (en) | 2023-06-12 | 2023-06-12 | Inertial navigation trajectory parameter measurement method based on Conv-LSTM and Kalman filtering fusion |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116538872A true CN116538872A (en) | 2023-08-04 |
Family
ID=87452574
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310692166.3A Pending CN116538872A (en) | 2023-06-12 | 2023-06-12 | Inertial navigation trajectory parameter measurement method based on Conv-LSTM and Kalman filtering fusion |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116538872A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117706583A (en) * | 2023-12-29 | 2024-03-15 | 无锡物联网创新中心有限公司 | High-precision positioning method and system |
-
2023
- 2023-06-12 CN CN202310692166.3A patent/CN116538872A/en active Pending
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117706583A (en) * | 2023-12-29 | 2024-03-15 | 无锡物联网创新中心有限公司 | High-precision positioning method and system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109596018B (en) | High-precision estimation method for flight attitude of spinning projectile based on magnetic roll angle rate information | |
| CN109373833B (en) | Combined measurement method suitable for initial attitude and speed of spinning projectile | |
| CN105929836B (en) | Control method for quadrotor | |
| CN109373832B (en) | Method for measuring initial parameters of rotating projectile muzzle based on magnetic rolling | |
| CN103017765B (en) | Yaw angle correction method and yaw angle correction device applied to a micro-mechanical integrated navigation system | |
| CN105180728B (en) | Front data based rapid air alignment method of rotary guided projectiles | |
| CN109724624B (en) | Airborne self-adaptive transfer alignment method suitable for wing deflection deformation | |
| CN105486308B (en) | Estimation plays the design method of the rapid convergence Kalman filter of line of sight angular speed | |
| CN116538872A (en) | Inertial navigation trajectory parameter measurement method based on Conv-LSTM and Kalman filtering fusion | |
| CN105115508A (en) | Post data-based rotary guided projectile quick air alignment method | |
| CN110044321B (en) | Method for resolving aircraft attitude by using geomagnetic information and angular rate gyroscope | |
| CN106979781A (en) | High-precision Transfer Alignment based on distributed inertance network | |
| CN105486307A (en) | Line-of-sight angular rate estimating method of maneuvering target | |
| CN113847913A (en) | Missile-borne integrated navigation method based on ballistic model constraint | |
| CN115248038B (en) | SINS/BDS combined navigation engineering algorithm under emission system | |
| CN111504256A (en) | Roll angle real-time estimation method based on least square method | |
| CN110220415B (en) | Closed-loop correction simulation platform and simulation method for outer trajectory of guided ammunition | |
| CN102853834A (en) | High-precision scheme of IMU for rotating carrier and denoising method | |
| CN112710298A (en) | Rotating missile geomagnetic satellite combined navigation method based on assistance of dynamic model | |
| CN109211232B (en) | Shell attitude estimation method based on least square filtering | |
| CN110160519A (en) | Body attitude calculation method for pulse shape modification rocket projectile | |
| CN113587740A (en) | Passive anti-radiation guiding method and system based on line-of-sight angle of bullet eyes | |
| Wang et al. | Virtual gyros construction and evaluation method based on BILSTM | |
| CN114707317A (en) | Method and system for measuring flight parameters of spinning projectile based on trajectory prior knowledge | |
| CN112683265A (en) | MIMU/GPS integrated navigation method based on rapid ISS collective filtering |
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 |