CN113252031A - NARX neural network assisted integrated navigation method - Google Patents
NARX neural network assisted integrated navigation method Download PDFInfo
- Publication number
- CN113252031A CN113252031A CN202110469796.5A CN202110469796A CN113252031A CN 113252031 A CN113252031 A CN 113252031A CN 202110469796 A CN202110469796 A CN 202110469796A CN 113252031 A CN113252031 A CN 113252031A
- Authority
- CN
- China
- Prior art keywords
- neural network
- gps
- increment
- output
- sins
- 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
- 238000013528 artificial neural network Methods 0.000 title claims abstract description 71
- 238000000034 method Methods 0.000 title claims abstract description 27
- 238000005259 measurement Methods 0.000 claims abstract description 38
- 238000001914 filtration Methods 0.000 claims abstract description 30
- 238000012549 training Methods 0.000 claims abstract description 7
- 239000013598 vector Substances 0.000 claims description 21
- 230000001133 acceleration Effects 0.000 claims description 11
- 238000004364 calculation method Methods 0.000 claims description 7
- 230000005540 biological transmission Effects 0.000 claims description 6
- 238000003062 neural network model Methods 0.000 claims description 5
- 210000002569 neuron Anatomy 0.000 claims description 4
- 238000012886 linear function Methods 0.000 claims description 3
- 239000011159 matrix material Substances 0.000 description 11
- 230000006870 function Effects 0.000 description 8
- 238000010586 diagram Methods 0.000 description 3
- 238000002474 experimental method Methods 0.000 description 2
- 230000007774 longterm Effects 0.000 description 2
- 238000005295 random walk Methods 0.000 description 2
- 230000007704 transition Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000000295 complement effect Effects 0.000 description 1
- 238000013329 compounding Methods 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 238000013178 mathematical model Methods 0.000 description 1
- 238000013441 quality evaluation Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 230000006403 short-term memory Effects 0.000 description 1
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
- 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
- G01S19/50—Determining position whereby the position solution is constrained to lie upon a particular curve or surface, e.g. for locomotives on railway tracks
-
- 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
-
- 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
- G06N3/08—Learning methods
-
- 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
Abstract
The invention discloses a method for NARX neural network assisted integrated navigation, which relates to the technical field of navigation and comprises the following contents: during the normal work of the GPS, the neural network receives IMU output information and SINS position speed increment as neural network input, and the GPS position speed increment is used as network target output to carry out neural network training; and in the short-time failure period of the GPS, the NARX neural network gives a predicted value of the GPS position speed increment according to the input IMU information and the SINS position speed increment, and the predicted value and the SINS position speed increment are subjected to difference to obtain the quantity measurement required by Kalman filtering. The invention ensures the smooth measurement updating and counteracts the influence of short-time failure of the GPS on the integrated navigation precision.
Description
Technical Field
The invention mainly relates to the technical field of navigation, in particular to a method for NARX neural network assisted integrated navigation.
Background
Any navigation mode has advantages and disadvantages, and the navigation requirement of the vehicle is difficult to meet by using one navigation mode alone. The integrated navigation is to combine multiple navigation systems with complementary advantages and disadvantages, measure and calculate the same navigation information to form a measurement value, calculate the error of each navigation system from the measurement value and correct the error, thereby improving the navigation precision. Combined Navigation systems combining Strap-down Inertial Navigation (SINS) and Global Positioning System (GPS) are widely used in various vehicles. The SINS does not need any external information or radiate any information outwards, can normally work under any medium and environment conditions, and can output rich navigation parameters including carrier attitude, speed and position, and has high short-time navigation precision but accumulated long-time navigation errors, so that the navigation errors are increased along with the navigation duration, and the long-term stability is poor. The GPS navigation precision is not influenced by navigation time length, the long-term stability is good, the GPS navigation precision and the SINS can be corrected when combined, but the GPS completely depends on navigation information sent by a navigation satellite, satellite signals are easy to be interfered or shielded, and the GPS short-time failure condition occurs.
Artificial Neural Networks (ANNs) are mathematical models inspired by biology and neuroscience, are large-scale parallel distributed processors, and naturally have the ability to store and use empirical knowledge. The feed-Forward Neural Network (FNN) is the earliest invented simple ANN, and FNN can be regarded as a function, and nonlinear mapping of input to output is realized by compounding simple nonlinear functions. The neural network of Nonlinear Auto-Regressive with Exogenous Inputs (NARX) is a recursive dynamic neural network, has short-term memory capability, can be regarded as FNN with input delay, output feedback and output feedback delay, and can better learn the relationship between the input and the output of a complex dynamic system. The method is applied to a combined navigation system, and can better give a predicted value of the GPS position and speed information according to IMU output information and inertial navigation resolving information.
When a vehicle runs in an area with GPS signal shielding or interference, the condition of GPS failure is easy to occur, so that the combined navigation filter cannot be measured and updated, and further the combined navigation precision is rapidly reduced in a short time. When the carrier runs in a tunnel, an overpass, a high-rise forest and a luxuriant area with trees, GPS signals in SINS/GPS integrated navigation are interfered or shielded, GPS failure is caused, required position and speed observation quantity cannot be provided for Kalman filtering, filtering measurement updating cannot be normally carried out due to lack of observation quantity, an integrated navigation system is equal to a system working in a pure inertial navigation state, and navigation errors can be accumulated along with navigation duration.
Disclosure of Invention
The technical problem to be solved by the invention is to provide a method for NARX neural network assisted integrated navigation, which effectively solves the problem of navigation precision of an SINS/GPS integrated navigation system caused by short-time GPS failure and counteracts the influence of the short-time GPS failure on the integrated navigation precision.
In order to solve the technical problems, the technical scheme adopted by the invention is as follows:
a method of NARX neural network assisted integrated navigation, comprising the steps of:
step 1: constructing a NARX neural network model;
step 2: carrying out inertial navigation calculation according to the 3-axis angular velocity and the 3-axis acceleration output by the inertial measurement unit to obtain an east position and velocity increment and a north position and velocity increment;
and step 3: during the normal working period of the GPS, training the NARX neural network constructed in the step 1, wherein the training data come from an inertial measurement unit, inertial navigation resolving output and the position and speed of GPS output;
and 4, step 4: during the short-time failure of the GPS, the NARX neural network trained in the step 3 gives a predicted value of the GPS position and speed increment according to the inertial measurement unit data and the inertial navigation resolving data;
and 5: and (3) subtracting the GPS position increment predicted value output by the NARX neural network in the step (4) from the position increment obtained by inertial navigation solution to obtain the position measurement required by measurement update of Kalman filtering, and subtracting the GPS speed increment predicted value output by the NARX neural network in the step (4) from the speed increment obtained by inertial navigation solution to obtain the speed measurement required by measurement update of the Kalman filtering, so that the normal operation of the Kalman filtering is ensured.
The technical scheme of the invention is further improved as follows: the input-output relationship of the NARX neural network model constructed in the step 1 is as follows:
y(t)＝f)x(t),x(t-1),y(t-1))，
wherein t represents the current time, t-1 represents the previous time, f (x, y) is a multivariate nonlinear function with respect to x, y (t), y (t-1) represent the current time and the previous time network output, x (t), x (t-1) represent the current time and the previous time network input, the neural network output depends on the input at the current time, the input at the previous time, the output at the previous time; the neural network comprises an input layer, a hidden layer and an output layer, wherein the hidden layer comprises 10 neurons, the transmission function of the hidden layer is tan sig, and the transmission function of the output layer is a linear function.
The technical scheme of the invention is further improved as follows: in step 3, the 3-axis angular velocity and the 3-axis acceleration output by the IMU, the east position and velocity increment and the north position and velocity increment obtained by inertial navigation calculation in step 2 are selected as neural network inputs, the east position and velocity increment and the north position and velocity increment of the GPS are selected as neural network target outputs, and the neural network is trained.
The technical scheme of the invention is further improved as follows: in order to obtain the observations required by the kalman filtering in step 5, namely the east position and velocity difference and the north position and velocity difference of the SINS and the GPS, the east position of the GPS output by the NARX neural network in step 4 needs to be incremented (Δ Pe)GNSS) Northbound position increment (Δ Pn)GNSS) East speed delta (Δ Ve)GNSS) Northbound speed increment (Δ Vn)GNSS) 4 terms, respectively, and east position increment (delta Pe) obtained by inertial navigationSINS) Northbound position increment (Δ Pn)SINS) East speed delta (Δ Ve)SINS) Northbound speed increment (Δ Vn)SINS) Subtracting 4 corresponding items to obtain 4 observed quantities required by measurement updating of Kalman filtering, and ensuring the normality of Kalman filteringIs carried out according to the formula
Wherein δ P (t), δ V (t) are east and north position error vectors, east and north velocity error vectors at time t, i.e. observed quantities required for measurement and update by Kalman filtering, PSINS(t)、VSINS(t) INS position vector and velocity vector, P, respectively at time tGNSS(t)、VGNSS(t) GPS position vector and velocity vector, Δ P, at time t, respectivelySINS(t) and Δ VSINS(t) SINS position and velocity increments, P, respectively, at time tt-1、Vt-1To combine navigation position and velocity, Δ P, for time t-1, respectivelyGNSS(t) and Δ VGNSSAnd (t) respectively giving predicted values of the GPS position increment and the speed increment by the neural network in the step 4 at the time t. Due to the adoption of the technical scheme, the invention has the technical progress that:
the method effectively solves the problem caused by short-time GPS failure on the navigation precision of the SINS/GPS integrated navigation system, counteracts the influence of the short-time GPS failure on the navigation precision of the SINS/GPS integrated navigation system, learns the complex dynamic relation between IMU output information and inertial navigation resolving information and GNSS position speed information through the NARX neural network, and gives a GPS position speed increment predicted value for calculating the observed quantity required by Kalman filtering during the short-time GPS failure so as to ensure the normal running of Kalman filtering measurement updating.
Drawings
FIG. 1 is a simplified NARX neural network architecture;
FIG. 2 is a schematic diagram of a neural network aiding principle;
FIG. 3 is a schematic diagram of a combined navigation output motion trajectory;
FIG. 4 is a schematic diagram of the combined navigation output position error.
Detailed Description
The present invention will be described in further detail with reference to the following examples:
as shown in fig. 1, fig. 2, fig. 3, and fig. 4, a method for NARX neural network assisted integrated navigation aims to counteract the influence of short-term GPS failure on the navigation accuracy of an SINS/GPS integrated navigation system, learns the complex dynamic relationship between IMU output information and inertial navigation solution information and GNSS position speed information through the NARX neural network, and provides a GPS position speed increment prediction value for calculating the observed quantity required by kalman filtering during the GPS short-term failure by the NARX neural network, thereby ensuring the normal operation of kalman filtering measurement update.
The invention relates to a method for NARX neural network assisted integrated navigation, which specifically comprises the following steps:
step 1: the method comprises the steps of constructing an NARX neural network model, wherein a simplified NARX neural network structure is shown in figure 1, the constructed model comprises an input layer, a hidden layer and an output layer, the hidden layer comprises 10 neurons, a transmission function is tansig, the input of the neural network is 10-dimensional information, the output of the neural network is 4-dimensional information, the transmission function of the neurons of the output layer is a linear function, and the input delay number and the output feedback delay number of the neural network are both 1.
The neural network output depends on the input at the current time, the input at the previous time, and the output at the previous time, and the input-output relationship can be represented by the following formula:
y)t)＝f(x(t),x(t-1),y(t-1))，
wherein t represents the current time, t-1 represents the previous time, f (x, y) is a multivariate nonlinear function with respect to x, y (t), y (t-1) represent the current time and the previous time network output, and x (t), x (t-1) represent the current time and the previous time network input.
Step 2: the method comprises the steps of carrying out inertial navigation calculation according to 3-axis angular velocity and 3-axis acceleration output by an Inertial Measurement Unit (IMU) to obtain an east position and velocity increment, a north position and velocity increment, wherein the IMU comprises a 3-axis gyroscope and a 3-axis accelerometer, the gyroscope is used for measuring angular motion information, the accelerometer is used for measuring linear motion information, and the IMU generally outputs the 3-axis angular velocity increment and the 3-axis velocity increment. The formula on which the inertial navigation is solved is
In the formula VnIs a three-dimensional velocity vector under a navigation coordinate system,
Is an attitude matrix, fbOutputting specific force information for IMU,
Coriolis accelerations caused by the motion of the vehicle relative to the earth and the rotation of the earth and centripetal acceleration caused by the motion of the vehicle on the earth's surface, g is gravitational acceleration, VN, VE are north and east velocities, RM, RN are radii of curvature of the meridian and the prime circle of the vehicle, and L, h is the latitude and altitude of the location of the vehicle.
The SINS/GPS integrated navigation mode adopts a loose integration mode, the integrated navigation filtering method is Kalman filtering, and the Kalman filtering state vector is
The method comprises a 3-axis misalignment angle, a 3-dimensional speed error, a 3-dimensional position error, a 3-axis gyroscope zero offset and a 3-axis accelerometer zero offset, wherein Kalman filtering observation vectors comprise a 3-dimensional position and a 3-dimensional speed error.
Z＝[Psins-Pgnss，Vsins-Vgnss]，
Kalman filtering equation of state and measurement equation of
Xk＝Φk，k-1XR-1+Гk-1WR-1，
Zk＝HkXk+Vk,
Phi is a state transition matrix which can be obtained according to an inertial navigation error equation, gamma is a system noise driving matrix, W is a system noise matrix, H is a measurement matrix, and V is a measurement noise matrix. The state transition matrix, the noise driving matrix and the system noise matrix contain system dynamic characteristic related information and are used for evaluating the quality of the system state estimation predicted value. The measurement matrix represents the relationship between the measurement information and each state quantity, and the measurement noise matrix comprises measurement noise characteristics for evaluating the quality of the measurement information. And (4) integrating the system state estimation predicted value and the quality evaluation of the measurement information, and determining the utilization ratio of the measurement value and the state estimation predicted value.
And step 3: during the normal work of the GPS, the neural network is in a training state, as shown in figure 2 (working with solid lines and dotted lines, invalid), the NARX neural network constructed in the step 1 is trained, the neural network inputs 3-axis angular velocity and 3-axis acceleration output by the IMU, the east position and velocity increment and the north position and velocity increment obtained by inertial navigation in the step 2, and the neural network target outputs the east position and velocity increment and the north position and velocity increment of the GPS. The neural network training method adopts a Levenberg-Marquardt method.
And 4, step 4: during the GPS failure, the neural network is switched to a prediction state, as shown in the attached figure 2 (working in a solid line and a dotted line, and invalid in a dotted line), the NARX neural network trained in the step 3 gives predicted values of the east position and the velocity increment and the north position and the velocity increment of the GPS according to the input 3-axis angular velocity, the 3-axis acceleration, the east position and the velocity increment obtained by resolving through inertial navigation and the north position and the velocity increment.
And 5: obtaining observations required for Kalman filtering, namely east position and speed difference and north position and speed difference of SINS and GPSThe GPS east position output by the NARX neural network in step 4 needs to be incremented by (Δ Pe)GNSS) Northbound position increment (Δ Pn)GNSS) East speed delta (Δ Ve)GNSS) Northbound speed increment (Δ Vn)GNSS) 4 terms, respectively, and east position increment (delta Pe) obtained by inertial navigationSINS) Northbound position increment (Δ Pn)SINS) East speed delta (Δ Ve)SINS) Northbound speed increment (Δ Vn)SINS) And subtracting 4 corresponding items to obtain 4 observed quantities required by measurement updating of Kalman filtering, so as to ensure the normal operation of the Kalman filtering according to a formula
δ P (t) and δ V (t) are east and north position error vectors and east and north speed error vectors at time t, PSINS (t) and VSINS (t) are SINS position vectors and speed vectors at time t, PGNSS (t) and VGNSS (t) are GPS position vectors and speed vectors at time t, Δ PSINS (t) and Δ VSINS (t) are INS position increment and speed increment at time t, Pt-1 and Vt-1 are combined navigation position and speed at time t-1, and Δ PGNSS (t) and Δ VGNSS (t) are predicted values of GPS position increment and speed increment given by the neural network in step 4 at time t.
Examples of the present invention are described below:
the experimental data is downloaded from www.psins.org.cn website by selecting a group of vehicle navigation data provided by Severe Engineer of northwest university of industry. The duration of the navigation data selected in the experiment is 1500 seconds. The model of the inertial measurement unit is STIM300, the zero offset stability of the gyroscope is 0.5 degrees/h, and the random walk of the gyroscope is 0.15 degrees/v/h (hours per degree). The accelerometer has a zero bias stability of 0.05mg and an acceleration random walk of (0.06m/s)/√ h (meters per second per hour). The IMU sampling frequency is 125Hz, and the single-point GNSS output frequency is 1 Hz. The total duration of the navigation data is 1500 seconds, and the duration from 1240 seconds to 1300 seconds is the simulated GNSS dead time period and is 60 seconds.
And the position and the speed which are solved by the normal work of the combined navigation system are used as the true value of the navigation information. And under the condition that the GPS fails, the position and the speed given by the neural network assisted navigation are experimental values of navigation information. And subtracting the true value from the experimental value to obtain a navigation error of the neural network assisted navigation, and judging the effectiveness of the neural network assisted integrated navigation method according to the error, wherein the smaller the navigation error is, the more stable the error curve is, and the stronger the capability of the neural network for counteracting the influence of the GPS failure on the integrated navigation is proved. The position information output at the initial stage of filtering has large fluctuation, which is not beneficial to clear display of the motion trail, and in order to highlight the difference of output trails of different methods in the GPS failure period, only the navigation trail of 550 seconds to 1500 seconds and the corresponding navigation error curve are given, and the combined navigation output motion trail is shown in figure 3.
And the solid line is a track output after GPS information and SINS information are fused by Kalman filtering under the normal working condition of the GPS, and the track is taken as an ideal motion track. The interval marked by the cross sign is the simulated GPS failure interval, and the time duration is 60 seconds. Before 1240 seconds, the GPS works normally, the combined navigation system is not interfered, and the motion tracks output by different methods are basically overlapped. The 1240 seconds to 1300 seconds are GPS failure intervals, the motion tracks output by different methods have larger difference, and the dotted line is the motion track output under the pure inertial navigation calculation mode and has obvious deviation with the ideal motion track. The dotted line is a navigation track output by NARX neural network assisted navigation, and the output track of NARX neural network assisted navigation is obviously closer to an ideal motion track.
The combined navigation output position error is shown in fig. 4, during the GPS failure, the root mean square error of the east position is 22.5 meters in the pure inertial navigation calculation mode, and the root mean square error of the north position is 22 meters; under the assistance of the NARX neural network, the root mean square error of the east position is 1.8 meters, and the root mean square error of the north position is 1.1 meters. Compared with a pure inertial navigation mode, the navigation error of the NARX neural network assisted combined navigation method is reduced by about 90%. Experiments prove that the combined navigation method assisted by the NARX neural network can better reduce the influence of short-time GPS failure on the combined navigation precision.
Claims (4)
1. A method for NARX neural network assisted integrated navigation is characterized in that: the method comprises the following steps:
step 1: constructing a NARX neural network model;
step 2: carrying out inertial navigation calculation according to the 3-axis angular velocity and the 3-axis acceleration output by the inertial measurement unit to obtain an east position and velocity increment and a north position and velocity increment;
and step 3: during the normal working period of the GPS, training the NARX neural network constructed in the step 1, wherein the training data come from an inertial measurement unit, inertial navigation resolving output and the position and speed of GPS output;
and 4, step 4: during the short-time failure of the GPS, the NARX neural network trained in the step 3 gives a predicted value of the GPS position and speed increment according to the inertial measurement unit data and the inertial navigation resolving data;
and 5: and (3) subtracting the GPS position increment predicted value output by the NARX neural network in the step (4) from the position increment obtained by inertial navigation solution to obtain the position measurement required by measurement update of Kalman filtering, and subtracting the GPS speed increment predicted value output by the NARX neural network in the step (4) from the speed increment obtained by inertial navigation solution to obtain the speed measurement required by measurement update of the Kalman filtering, so that the normal operation of the Kalman filtering is ensured.
2. The method of NARX neural network assisted integrated navigation of claim 1, wherein: the input-output relationship of the NARX neural network model constructed in the step 1 is as follows:
y(t)＝f(x(t),x(t-1),y(t-1))，
wherein t represents the current time, t-1 represents the previous time, f (x, y) is a multivariate nonlinear function with respect to x, y (t), y (t-1) represent the current time and the previous time network output, x (t), x (t-1) represent the current time and the previous time network input, the neural network output depends on the input at the current time, the input at the previous time, the output at the previous time; the neural network comprises an input layer, a hidden layer and an output layer, wherein the hidden layer comprises 10 neurons, the transmission function of the hidden layer is tan sig, and the transmission function of the output layer is a linear function.
3. The method of NARX neural network assisted integrated navigation of claim 1, wherein: in step 3, the 3-axis angular velocity and the 3-axis acceleration output by the IMU, the east position and velocity increment and the north position and velocity increment obtained by inertial navigation calculation in step 2 are selected as neural network inputs, the east position and velocity increment and the north position and velocity increment of the GPS are selected as neural network target outputs, and the neural network is trained.
4. The method of NARX neural network assisted integrated navigation of claim 1, wherein: in order to obtain the observations required by the kalman filtering in step 5, namely the east position and velocity difference and the north position and velocity difference of the SINS and the GPS, the east position of the GPS output by the NARX neural network in step 4 needs to be incremented (Δ Pe)GNSS) Northbound position increment (Δ Pn)GNSS) East speed delta (Δ Ve)GNSS) Northbound speed increment (Δ Vn)GNSS) 4 terms, respectively, and east position increment (delta Pe) obtained by inertial navigationSINS) Northbound position increment (Δ Pn)SINS) East speed delta (Δ Ve)SINS) Northbound speed increment (Δ Vn)SINS) And subtracting 4 corresponding items to obtain 4 observed quantities required by measurement updating of Kalman filtering, so as to ensure the normal operation of the Kalman filtering according to a formula
Wherein δ P (t), δ V (t) are east and north position error vectors, east and north velocity error vectors at time t, i.e. observed quantities required for measurement and update by Kalman filtering, PSINS(t)、VSINS(t) INS position vector and velocity vector, P, respectively at time tGNSS(t)、VGNSS(t) GPS position vector and velocity vector, Δ P, at time t, respectivelySINS(t) and Δ VSINS(t) SINS position and velocity increments, P, respectively, at time tt-1、Vt-1To combine navigation position and velocity, Δ P, for time t-1, respectivelyGNSS(t) and Δ VGNSSAnd (t) respectively giving predicted values of the GPS position increment and the speed increment by the neural network in the step 4 at the time t.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110469796.5A CN113252031A (en) | 2021-04-28 | 2021-04-28 | NARX neural network assisted integrated navigation method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110469796.5A CN113252031A (en) | 2021-04-28 | 2021-04-28 | NARX neural network assisted integrated navigation method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113252031A true CN113252031A (en) | 2021-08-13 |
Family
ID=77222222
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110469796.5A Pending CN113252031A (en) | 2021-04-28 | 2021-04-28 | NARX neural network assisted integrated navigation method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113252031A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113847915A (en) * | 2021-09-24 | 2021-12-28 | 中国人民解放军战略支援部队信息工程大学 | Navigation method of strapdown inertial navigation/Doppler integrated navigation system |
| CN114593735A (en) * | 2022-01-26 | 2022-06-07 | 奥比中光科技集团股份有限公司 | Pose prediction method and device |
| CN114690221A (en) * | 2021-12-22 | 2022-07-01 | 北京航天时代激光导航技术有限责任公司 | Elman neural network and Kalman fusion filtering algorithm based on wavelet threshold method |
| CN115143954A (en) * | 2022-09-05 | 2022-10-04 | 中国电子科技集团公司第二十八研究所 | Unmanned vehicle navigation method based on multi-source information fusion |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104898148A (en) * | 2015-06-02 | 2015-09-09 | 北京航空航天大学 | Low-cost INS/GPS seamless navigation method based on data compression and neural network |
| CN108957510A (en) * | 2018-07-25 | 2018-12-07 | 南京航空航天大学 | Based on inertia/zero-speed/GPS pedestrian is seamless combined navigation locating method |
| CN109521454A (en) * | 2018-12-06 | 2019-03-26 | 中北大学 | A kind of GPS/INS Combinated navigation method based on self study volume Kalman filtering |
| CN109781099A (en) * | 2019-03-08 | 2019-05-21 | 兰州交通大学 | A kind of navigation methods and systems of adaptive UKF algorithm |
| US20200355503A1 (en) * | 2018-01-10 | 2020-11-12 | Oxford University Innovation Limited | Determining the location of a mobile device |
| CN112505737A (en) * | 2020-11-16 | 2021-03-16 | 东南大学 | GNSS/INS combined navigation method based on Elman neural network online learning assistance |
-
2021
- 2021-04-28 CN CN202110469796.5A patent/CN113252031A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104898148A (en) * | 2015-06-02 | 2015-09-09 | 北京航空航天大学 | Low-cost INS/GPS seamless navigation method based on data compression and neural network |
| US20200355503A1 (en) * | 2018-01-10 | 2020-11-12 | Oxford University Innovation Limited | Determining the location of a mobile device |
| CN108957510A (en) * | 2018-07-25 | 2018-12-07 | 南京航空航天大学 | Based on inertia/zero-speed/GPS pedestrian is seamless combined navigation locating method |
| CN109521454A (en) * | 2018-12-06 | 2019-03-26 | 中北大学 | A kind of GPS/INS Combinated navigation method based on self study volume Kalman filtering |
| CN109781099A (en) * | 2019-03-08 | 2019-05-21 | 兰州交通大学 | A kind of navigation methods and systems of adaptive UKF algorithm |
| CN112505737A (en) * | 2020-11-16 | 2021-03-16 | 东南大学 | GNSS/INS combined navigation method based on Elman neural network online learning assistance |
Non-Patent Citations (2)
| Title |
|---|
| 方伟: "GNSS中断环境下融合导航算法关键技术研究", 《中国优秀博硕士学位论文数据库(硕士)信息科技辑》 * |
| 邓天民，等: "基于非线性自适应回归神经网络的GPS/IMU组合导航方法", 《科学技术与工程》 * |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113847915A (en) * | 2021-09-24 | 2021-12-28 | 中国人民解放军战略支援部队信息工程大学 | Navigation method of strapdown inertial navigation/Doppler integrated navigation system |
| CN113847915B (en) * | 2021-09-24 | 2023-12-19 | 中国人民解放军战略支援部队信息工程大学 | Navigation method of strapdown inertial navigation/Doppler integrated navigation system |
| CN114690221A (en) * | 2021-12-22 | 2022-07-01 | 北京航天时代激光导航技术有限责任公司 | Elman neural network and Kalman fusion filtering algorithm based on wavelet threshold method |
| CN114593735A (en) * | 2022-01-26 | 2022-06-07 | 奥比中光科技集团股份有限公司 | Pose prediction method and device |
| CN115143954A (en) * | 2022-09-05 | 2022-10-04 | 中国电子科技集团公司第二十八研究所 | Unmanned vehicle navigation method based on multi-source information fusion |
| CN115143954B (en) * | 2022-09-05 | 2022-11-29 | 中国电子科技集团公司第二十八研究所 | Unmanned vehicle navigation method based on multi-source information fusion |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113252031A (en) | NARX neural network assisted integrated navigation method | |
| Bevly | Global positioning system (GPS): A low-cost velocity sensor for correcting inertial sensor errors on ground vehicles | |
| JP5328252B2 (en) | Position detection apparatus and position detection method for navigation system | |
| CN103235328B (en) | GNSS (global navigation satellite system) and MEMS (micro-electromechanical systems) integrated navigation method | |
| CN109000640B (en) | Vehicle GNSS/INS integrated navigation method based on discrete grey neural network model | |
| CN102252677A (en) | Time series analysis-based variable proportion self-adaptive federal filtering method | |
| CN102818567A (en) | AUV (autonomous underwater vehicle) integrated navigation method integrating Kalman filtering and particle filtering | |
| CN103076026B (en) | A kind of method determining Doppler log range rate error in SINS | |
| JP5164645B2 (en) | Method and apparatus for repetitive calculation control in Kalman filter processing | |
| CN110440830A (en) | Vehicle-mounted Strapdown Inertial Navigation System Alignment Method under moving base | |
| CN107289942A (en) | A kind of relative navigation system and method for formation flight | |
| CN111323012A (en) | INS-assisted DVL speed measurement error compensation method under high dynamic environment of carrier | |
| CN110849360A (en) | Distributed relative navigation method for multi-machine cooperative formation flight | |
| Gao et al. | An integrated land vehicle navigation system based on context awareness | |
| CN117053782A (en) | Combined navigation method for amphibious robot | |
| CN111006675A (en) | Self-calibration method of vehicle-mounted laser inertial navigation system based on high-precision gravity model | |
| Pan et al. | Real-time accurate odometer velocity estimation aided by accelerometers | |
| CN116224407B (en) | GNSS and INS integrated navigation positioning method and system | |
| CN113008229A (en) | Distributed autonomous integrated navigation method based on low-cost vehicle-mounted sensor | |
| CN114111840B (en) | DVL error parameter online calibration method based on integrated navigation | |
| Belhajem et al. | A hybrid low cost approach using extended Kalman filter and neural networks for real time positioning | |
| RU2502049C1 (en) | Small-size platformless inertial navigation system of medium accuracy, corrected from system of air signals | |
| CN113847915B (en) | Navigation method of strapdown inertial navigation/Doppler integrated navigation system | |
| CN113551669B (en) | Combined navigation positioning method and device based on short base line | |
| Malleswaran et al. | A hybrid approach for GPS/INS integration using Kalman filter and IDNN |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20210813 |