CN117906601A - Multi-source sensor fusion-oriented adaptive interactive navigation positioning filtering method - Google Patents
Multi-source sensor fusion-oriented adaptive interactive navigation positioning filtering method Download PDFInfo
- Publication number
- CN117906601A CN117906601A CN202311644838.XA CN202311644838A CN117906601A CN 117906601 A CN117906601 A CN 117906601A CN 202311644838 A CN202311644838 A CN 202311644838A CN 117906601 A CN117906601 A CN 117906601A
- Authority
- CN
- China
- Prior art keywords
- model
- filtering
- filtering model
- output state
- state
- 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.)
- Granted
Links
- 238000001914 filtration Methods 0.000 title claims abstract description 221
- 238000000034 method Methods 0.000 title claims abstract description 37
- 230000002452 interceptive effect Effects 0.000 title claims abstract description 17
- 230000003044 adaptive effect Effects 0.000 title claims description 13
- 230000004927 fusion Effects 0.000 claims abstract description 58
- 230000003213 activating effect Effects 0.000 claims abstract description 18
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 17
- 239000011159 matrix material Substances 0.000 claims description 77
- 239000013598 vector Substances 0.000 claims description 9
- 230000008569 process Effects 0.000 claims description 6
- 230000007704 transition Effects 0.000 claims description 3
- 230000004048 modification Effects 0.000 description 6
- 238000012986 modification Methods 0.000 description 6
- 239000002245 particle Substances 0.000 description 4
- 238000012545 processing Methods 0.000 description 4
- 230000009466 transformation Effects 0.000 description 3
- 230000009286 beneficial effect Effects 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 230000035945 sensitivity Effects 0.000 description 2
- 238000000342 Monte Carlo simulation Methods 0.000 description 1
- 230000004913 activation Effects 0.000 description 1
- 230000006978 adaptation Effects 0.000 description 1
- 230000004075 alteration Effects 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000012512 characterization method Methods 0.000 description 1
- 230000001934 delay Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000009499 grossing Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000003909 pattern recognition Methods 0.000 description 1
- 230000010287 polarization Effects 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
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
- G01C21/20—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)
- Radar Systems Or Details Thereof (AREA)
Abstract
A self-adaptive interactive navigation positioning filtering method for multi-source sensor fusion comprises the following steps: constructing a fusion filtering model which comprises a first filtering model and a second filtering model and runs in parallel; acquiring initial information of a target object; according to a preset rule, alternately activating the first filtering model and the second filtering model to a preset number of times in one period; respectively calculating a first output state and a second output state according to a preset first algorithm; and carrying out weighted fusion to obtain the final output state of the fusion filtering model. According to the invention, the first filtering model and the second filtering model which run in parallel are set as the fusion filtering model, and the first filtering model and the second filtering model are alternately activated again to output the optimal estimation state.
Description
Technical Field
The invention relates to the technical field of navigation positioning, in particular to a multi-source sensor fusion-oriented self-adaptive interactive navigation positioning filtering method.
Background
The multi-source sensor fusion refers to that multiple sources, multiple forms, multiple angles, multiple layers, multiple scales and other modes such as different sensors, different channels, different time periods, different spatial positions, different frequency ranges, different polarization modes, different phase modes and the like are obtained in multiple modes such as the same target or the related information of a scene or a phenomenon waiting study object to be effectively integrated so as to obtain more complete, accurate, reliable, valuable, meaningful and other higher-quality information, so that the targets such as understanding, judging, deciding and the like of the study object are better realized. The multi-source sensor fusion technology is a technology for comprehensively utilizing multi-disciplinary knowledge and technology such as signal processing, image processing, pattern recognition, artificial intelligence, control theory, mathematical statistics and the like to solve the actual problem and obtain better effects and performances.
Disclosure of Invention
Object of the invention
The invention aims to provide a multi-source sensor fusion-oriented adaptive interactive navigation positioning filtering method which overcomes the limitation of the existing single filtering model and cannot adapt to complex sensor fusion scenes.
(II) technical scheme
In order to solve the problems, the invention provides a multi-source sensor fusion-oriented adaptive interactive navigation positioning filtering method, which comprises the following steps:
constructing a fusion filtering model, wherein the fusion filtering model comprises a first filtering model and a second filtering model, and the first filtering model and the second filtering model run in parallel;
Acquiring initial information of a target object in a period;
according to the initial information of the target object and a preset rule, alternately activating the first filtering model and the second filtering model to a preset number of times in one period;
respectively calculating a first output state of a first filtering model and a second output state of a second filtering model in a period according to a preset first algorithm;
And carrying out weighted fusion on the first output state and the second output state to obtain the final output state of the fusion filtering model.
In another aspect of the present invention, the initial information of the target object preferably includes observation data of the multi-source sensor and error feature information of the system.
In another aspect of the present invention, preferably, according to a preset rule according to the initial information of the target object, the alternately activating the first filtering model and the second filtering model for a preset number of times in one period includes:
If the error characteristic information of the system is available, activating a first filtering model at the current moment in one period, and activating a second filtering model at the next moment in the same period;
And if the error characteristic information of the system is not available, activating the second filtering model at the current moment in one period, and activating the first filtering model at the next moment in the same period.
In another aspect of the present invention, preferably, the first filtering model includes a plurality of sub-filtering models; the second filtering model includes a number of sub-filtering models.
In another aspect of the present invention, preferably, calculating the first output state of the first filter model and the second output state of the second filter model in one period according to a preset first algorithm includes:
Inputting the initial information into the activated first filtering model or the second filtering model;
predicting and updating each sub-filtering model in the activated first filtering model or the activated second filtering model to obtain the output state of each sub-filtering model;
calculating the mixed probability and model probability of each sub-filter model in the activated first filter model or the activated second filter model;
Carrying out state fusion according to the output state of each sub-filtering model in the activated first filtering model or the activated second filtering model and the model probability to obtain the first output state of the activated first filtering model or the second output state of the second filtering model;
And sending the first output state of the activated first filter model or the second output state of the second filter model to the fusion filter model, and taking the first output state or the second output state of the activated first filter model or the second output state of the second filter model as an initial value of the input state of each sub-filter model in the second filter model or the first filter model at the next moment according to a preset second algorithm.
In another aspect of the present invention, preferably, the performing weighted fusion on the first output state and the second output state to obtain an output state of a final fusion filtering model includes:
calculating a mahalanobis distance between the first output state and the second output state;
Determining a weighting coefficient of the first output state and the second output state according to the mahalanobis distance;
and linearly combining the first output state and the second output state according to the weighting coefficient to obtain the final output state of the fusion filtering model.
In another aspect of the present invention, preferably, the observation data of the multisource sensor is acquired by:
Acquiring target data of a multi-source sensor;
determining a sensor weight coefficient corresponding to each sensor;
Carrying out weighted average on target data of each sensor according to the corresponding sensor weight coefficient to obtain observation data of the multi-source sensor;
The error characteristic information of the system is obtained through the following steps:
determining a corresponding noise covariance matrix according to the type and the characteristics of the multi-source sensor;
Carrying out weighted average on the noise covariance matrix of each sensor according to the sensor weight coefficient to obtain a fused noise covariance matrix;
The noise covariance matrix is the error characteristic information of the system.
In another aspect of the present invention, preferably, the predicting and updating each sub-filtering model in the activated first filtering model or the activated second filtering model, to obtain an output state of each sub-filtering model includes:
Calculating the mixing state, the mixing state covariance matrix and the mixing model probability of each sub-filtering model according to the initial mixing probability, the initial model probability and the model transition probability matrix of each sub-filtering model;
calculating the prediction state, the prediction state covariance matrix and the prediction observation data of each sub-filtering model according to the mixed state, the mixed state covariance matrix and the system state equation;
calculating an update state, an update state covariance matrix, an update model probability and a likelihood function of each sub-filtering model according to the prediction state, the prediction state covariance matrix, the prediction observation data, the observation data and the observation equation;
And obtaining the output state and the output state covariance matrix of each sub-filtering model according to the update state, the update state covariance matrix, the update model probability and the likelihood function.
In another aspect of the present invention, preferably, the initial value of the input state of each sub-filtering model in the second filtering model or the first filtering model as the next moment corresponding to the preset second algorithm includes:
calculating a residual vector of each sub-filtering model, wherein the residual vector is the difference between the observed data and the predicted observed data;
And calculating a residual covariance matrix of each sub-filtering model, wherein the residual covariance matrix is the sum of a predicted observation covariance matrix and an observation noise covariance matrix.
In another aspect of the present invention, preferably, the initial value of the input state of each sub-filtering model in the second filtering model or the first filtering model corresponding to the preset second algorithm as the next moment further includes:
Calculating the average value and standard deviation of residual vectors;
Calculating the trace of the residual covariance matrix;
Determining a scaling factor based on the mean, standard deviation, and trace;
multiplying the scaling factor by the identity matrix to obtain an estimated observed noise covariance matrix;
the estimated observed noise covariance matrix is used as a new observed noise covariance matrix and is used for the filtering process at the next moment.
(III) beneficial effects
The technical scheme of the invention has the following beneficial technical effects:
According to the invention, the first filtering model or the second filtering model is activated according to the initial information received from the fusion filtering model and a preset rule, so that the first filtering model or the second filtering model outputs an optimal estimation state.
Drawings
FIG. 1 is an overall flow chart of one embodiment of the present invention;
Fig. 2 is an activation flow diagram according to one embodiment of the invention.
Detailed Description
The objects, technical solutions and advantages of the present invention will become more apparent by the following detailed description of the present invention with reference to the accompanying drawings. It should be understood that the description is only illustrative and is not intended to limit the scope of the invention. In addition, in the following description, descriptions of well-known structures and techniques are omitted so as not to unnecessarily obscure the present invention.
It will be apparent that the described embodiments are some, but not all, embodiments of the invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
In the description of the present invention, it should be noted that the terms "first," "second," and "third" are used for descriptive purposes only and are not to be construed as indicating or implying relative importance.
In addition, the technical features of the different embodiments of the present invention described below may be combined with each other as long as they do not collide with each other.
Examples
FIG. 1 shows an overall flowchart of one embodiment of the present invention, as shown in FIG. 1, comprising:
Constructing a fusion filtering model, wherein the fusion filtering model comprises a first filtering model and a second filtering model, and the first filtering model and the second filtering model run in parallel; the specific content of the first filtering model and the second filtering model is not limited herein, and the first filtering model and the second filtering model may be the same filtering model or may be different filtering models, and in this embodiment, the first filtering model and the second filtering model are different filtering models, but are both interactive multimode filtering models (IMM filtering models), and the specific content of the first filtering model and the second filtering model is not limited herein, and in this embodiment, the filtering models are Kalman Filtering (KF), unscented Kalman Filtering (KF) and Particle Filtering (PF), and in this embodiment, the first filtering model is unscented Kalman Filtering (KF) and the second filtering model is Particle Filtering (PF), where the UKF is a nonlinear filtering method based on unscented transformation, and can effectively process highly nonlinear system state equations and observation equations. The PF is a non-linear filtering method based on a Monte Carlo method, and can effectively process a system state equation and an observation equation which are distributed arbitrarily, and a Kalman Filter (KF) or an Unscented Kalman Filter (UKF) is used for processing a non-linear and non-Gaussian system, but the filtering models have certain limitations and assumption conditions and cannot adapt to complex sensor fusion scenes, for example, the KF needs to be linear, the UKF needs to be known as a noise covariance matrix in the system state equation and the observation equation, and when the conditions are not satisfied, estimation deviation and instability of the KF and the UKF can occur. Particle Filtering (PF), which can process any form of system state equation and observation equation, but requires a large amount of particles for sampling and updating, and has large calculation amount and poor real-time performance. Therefore, in the embodiment, by setting two IMM filter models running in parallel, the output states of the two IMM filter models are weighted and fused, and the complementarity of the two IMM filter models can be utilized, so that the robustness and the adaptability of fusion filtering are enhanced;
Acquiring initial information of a target object in a period; the method of acquiring the information is not limited, and the information may be optionally acquired through a sensor, further, in this embodiment, initial information is acquired through a multi-source sensor, and specific contents of the multi-source sensor are not limited herein, and may be a speed sensor, a temperature sensor and the like, and the contents of the multi-source sensor are converted into a uniform coordinate system and a data format, optionally, in this embodiment, observation data of the multi-source sensor includes a position, a speed, a direction and the like of a target, where the uniform coordinate system and the data format are a geographic coordinate system and a matrix form; the specific content of one period is not limited, and the period time of one positioning can be one time, or one day, one hour, one minute and the like; the specific content of the initial information of the target object is not limited, and optionally, in this embodiment, the initial information of the target object includes observation data of the multi-source sensor and error feature information of the system; the specific ways of obtaining the observation data and the error feature information of the system are not limited herein, and optionally, in this embodiment, the method for obtaining the observation data of the multi-source sensor includes: acquiring target data of a multi-source sensor; determining a sensor weight coefficient corresponding to each sensor; carrying out weighted average on target data of each sensor according to the corresponding sensor weight coefficient to obtain observation data of the multi-source sensor;
The method for acquiring the error characteristic information of the system comprises the following steps: determining a corresponding noise covariance matrix according to the type and the characteristics of the multi-source sensor; carrying out weighted average on the noise covariance matrix of each sensor according to the sensor weight coefficient to obtain a fused noise covariance matrix; the noise covariance matrix is the error characteristic information of the system;
According to the initial information of the target object and a preset rule, alternately activating the first filtering model and the second filtering model to a preset number of times in one period; the specific content of the preset rule is not limited herein, and optionally, in this embodiment, the preset rule includes: if the error characteristic information of the system is available, activating a first filtering model at the current moment in one period, and activating a second filtering model at the next moment in the same period; if the error characteristic information of the system is unavailable, activating a second filtering model at the current moment in one period, and activating a first filtering model at the next moment in the same period;
Error characterization information of the system is not specifically expressed as: when the noise covariance matrix in the system state equation and the observation equation is not available, the accurate value or reasonable estimated value of the noise covariance matrix cannot be obtained, or the noise covariance matrix cannot be determined due to time variation, in which case, a nonlinear filtering method capable of processing arbitrary distribution, namely PF, needs to be used; PF can represent the state distribution by generating a set of random samples, and predict and update these samples according to the system state equation and observation equation, thus get the optimal state estimation and covariance matrix; the PF does not need to know the specific value of the noise covariance matrix, but only needs to know what distribution the noise obeys;
The error characteristic information of the system can be specifically expressed as: when the noise covariance matrix in the system state equation and the observation equation is available, it means that an accurate value or a reasonable estimated value of the noise covariance matrix can be obtained, and the noise covariance matrix does not change or changes little with time. In this case, a nonlinear filtering method capable of handling high nonlinearity, namely UKF, may be used. UKF can represent state distribution by selecting a group of sample points called sigma points, and performing unscented transformation on the sample points according to a system state equation and an observation equation, so as to obtain an optimal state estimation and covariance matrix. UKF needs to know the specific value of the noise covariance matrix in order to perform unscented transformation;
In the application of multi-source sensor fusion, the design of the efficient fusion filtering method is greatly constrained, and particularly when the error characteristic information of the system is unavailable, the embodiment can automatically select a proper filtering model according to the availability of the error characteristic information of the system, so that the efficiency and the precision of fusion filtering are improved.
Respectively calculating a first output state of a first filtering model and a second output state of a second filtering model in a period according to a preset first algorithm; the specific content of the preset first algorithm is not limited herein, and optionally, in this embodiment, the first filtering model includes a plurality of sub-filtering models; the second filtering model comprises a plurality of sub-filtering models; the preset first algorithm comprises the following steps:
Inputting the initial information into the activated first filtering model or the second filtering model;
predicting and updating each sub-filtering model in the activated first filtering model or the activated second filtering model to obtain the output state of each sub-filtering model;
calculating the mixed probability and model probability of each sub-filter model in the activated first filter model or the activated second filter model;
Carrying out state fusion according to the output state of each sub-filtering model in the activated first filtering model or the activated second filtering model and the model probability to obtain the first output state of the activated first filtering model or the second output state of the second filtering model;
And sending the first output state of the activated first filter model or the second output state of the second filter model to the fusion filter model, and taking the first output state or the second output state of the activated first filter model or the second output state of the second filter model as an initial value of the input state of each sub-filter model in the second filter model or the first filter model at the next moment according to a preset second algorithm.
The method for obtaining the final output state of the fusion filtering model by performing weighted fusion on the first output state and the second output state is not limited herein, and optionally, in this embodiment, the specific content includes:
calculating a mahalanobis distance between the first output state and the second output state;
Determining a weighting coefficient of the first output state and the second output state according to the mahalanobis distance;
and linearly combining the first output state and the second output state according to the weighting coefficient to obtain the final output state of the fusion filtering model.
Further, in this embodiment, the predicting and updating each sub-filtering model in the activated first filtering model or the activated second filtering model to obtain the output state of each sub-filtering model includes:
Calculating the mixing state, the mixing state covariance matrix and the mixing model probability of each sub-filtering model according to the initial mixing probability, the initial model probability and the model transition probability matrix of each sub-filtering model;
calculating the prediction state, the prediction state covariance matrix and the prediction observation data of each sub-filtering model according to the mixed state, the mixed state covariance matrix and the system state equation;
calculating an update state, an update state covariance matrix, an update model probability and a likelihood function of each sub-filtering model according to the prediction state, the prediction state covariance matrix, the prediction observation data, the observation data and the observation equation;
And obtaining the output state and the output state covariance matrix of each sub-filtering model according to the update state, the update state covariance matrix, the update model probability and the likelihood function.
Further, in this embodiment, the preset second algorithm includes:
calculating a residual vector of each sub-filtering model, wherein the residual vector is the difference between the observed data and the predicted observed data; and calculating a residual covariance matrix of each sub-filtering model, wherein the residual covariance matrix is the sum of a predicted observation covariance matrix and an observation noise covariance matrix.
Further comprises: calculating the average value and standard deviation of residual vectors; calculating the trace of the residual covariance matrix; determining a scaling factor based on the mean, standard deviation, and trace; multiplying the scaling factor by the identity matrix to obtain an estimated observed noise covariance matrix; the estimated observed noise covariance matrix is used as a new observed noise covariance matrix and is used for the filtering process at the next moment. The efficiency and accuracy of the fusion filtering can be improved.
The embodiment solves the problem that in the application of multi-source sensor fusion, the design of the efficient fusion filtering method is greatly constrained, especially when the error characteristic information of the system is not available. The filter estimation algorithm switches between different filtering models through model probability adaptation, and the filter input is reinitialized in one period to realize interaction between the models. The present invention uses two parallel running filter models, including different filter models, and is alternately activated according to the information received from the fusion filter to improve the reliability and accuracy of the proposed detection solution. By utilizing residual information to carry out variance matching, the online updating of the observed noise covariance matrix can be realized, thereby improving the positioning performance. Compared with the method for observing the noise covariance matrix by using fixed or empirical values in the prior art, the method can adapt to noise changes of different sensors and different environments, and the self-adaptability and the accuracy of fusion filtering are enhanced. The output states of the two IMM filters are subjected to weighted fusion, so that the complementarity of the two IMM filters can be utilized, and the robustness and the adaptability of fusion filtering can be enhanced.
Compared with the method in the prior art that only the output state of a single filtering model is used as the fusion filtering result, the method can effectively eliminate estimation deviation and instability possibly existing in the single filtering model. For example, the UKF-IMM filter model may suffer from tracking delays and excessive smoothing when the target maneuvers; whereas the PF-IMM filter model may exhibit tracking jitter and excessive sensitivity. According to the invention, the output states of the two IMM filter models are linearly combined according to the weighting coefficients by calculating the mahalanobis distance between the output states of the two IMM filter models and determining the weighting coefficients of the output states of the two IMM filter models according to the mahalanobis distance, so that a final fusion filter result is obtained. Thus, the present invention can suppress tracking delay and jitter while maintaining tracking accuracy and sensitivity.
It is to be understood that the above-described embodiments of the present invention are merely illustrative of or explanation of the principles of the present invention and are in no way limiting of the invention. Accordingly, any modification, equivalent replacement, improvement, etc. made without departing from the spirit and scope of the present invention should be included in the scope of the present invention. Furthermore, the appended claims are intended to cover all such changes and modifications that fall within the scope and boundary of the appended claims, or equivalents of such scope and boundary.
The invention has been described above with reference to the embodiments thereof. These examples are for illustrative purposes only and are not intended to limit the scope of the present invention. The scope of the invention is defined by the appended claims and equivalents thereof. Various alternatives and modifications can be made by those skilled in the art without departing from the scope of the invention, and such alternatives and modifications are intended to fall within the scope of the invention.
Although embodiments of the present invention have been described in detail, it should be understood that various changes, substitutions, and alterations can be made hereto without departing from the spirit and scope of the invention.
It is apparent that the above examples are given by way of illustration only and are not limiting of the embodiments. Other variations or modifications of the above teachings will be apparent to those of ordinary skill in the art. It is not necessary here nor is it exhaustive of all embodiments. While still being apparent from variations or modifications that may be made by those skilled in the art are within the scope of the invention.
Claims (10)
1. A self-adaptive interactive navigation positioning filtering method for multi-source sensor fusion is characterized by comprising the following steps:
constructing a fusion filtering model, wherein the fusion filtering model comprises a first filtering model and a second filtering model, and the first filtering model and the second filtering model run in parallel;
Acquiring initial information of a target object in a period;
according to the initial information of the target object and a preset rule, alternately activating the first filtering model and the second filtering model to a preset number of times in one period;
respectively calculating a first output state of a first filtering model and a second output state of a second filtering model in a period according to a preset first algorithm;
And carrying out weighted fusion on the first output state and the second output state to obtain the final output state of the fusion filtering model.
2. The adaptive interactive navigation positioning filtering method for multi-source sensor fusion according to claim 1, wherein the initial information of the target object comprises observation data of the multi-source sensor and error characteristic information of the system.
3. The adaptive interactive navigation positioning filtering method for multi-source sensor fusion according to claim 2, wherein alternately activating the first filtering model and the second filtering model to a preset number of times in one period according to a preset rule according to initial information of the target object comprises:
If the error characteristic information of the system is available, activating a first filtering model at the current moment in one period, and activating a second filtering model at the next moment in the same period;
And if the error characteristic information of the system is not available, activating the second filtering model at the current moment in one period, and activating the first filtering model at the next moment in the same period.
4. The adaptive interactive navigation positioning filtering method for multi-source sensor fusion according to claim 1, wherein the first filtering model comprises a plurality of sub-filtering models; the second filtering model includes a number of sub-filtering models.
5. The adaptive interactive navigation positioning filtering method for multi-source sensor fusion according to claim 4, wherein calculating the first output state of the first filtering model and the second output state of the second filtering model in one period according to a preset first algorithm respectively comprises:
Inputting the initial information into the activated first filtering model or the second filtering model;
predicting and updating each sub-filtering model in the activated first filtering model or the activated second filtering model to obtain the output state of each sub-filtering model;
calculating the mixed probability and model probability of each sub-filter model in the activated first filter model or the activated second filter model;
Carrying out state fusion according to the output state of each sub-filtering model in the activated first filtering model or the activated second filtering model and the model probability to obtain the first output state of the activated first filtering model or the second output state of the second filtering model;
And sending the first output state of the activated first filter model or the second output state of the second filter model to the fusion filter model, and taking the first output state or the second output state of the activated first filter model or the second output state of the second filter model as an initial value of the input state of each sub-filter model in the second filter model or the first filter model at the next moment according to a preset second algorithm.
6. The adaptive interactive navigation positioning filtering method for multi-source sensor fusion according to claim 1, wherein the weighted fusion of the first output state and the second output state to obtain the final output state of the fusion filtering model comprises:
calculating a mahalanobis distance between the first output state and the second output state;
Determining a weighting coefficient of the first output state and the second output state according to the mahalanobis distance;
and linearly combining the first output state and the second output state according to the weighting coefficient to obtain the final output state of the fusion filtering model.
7. The adaptive interactive navigation positioning filtering method for multi-source sensor fusion according to claim 2, wherein the observation data of the multi-source sensor is obtained by the following steps:
Acquiring target data of a multi-source sensor;
determining a sensor weight coefficient corresponding to each sensor;
Carrying out weighted average on target data of each sensor according to the corresponding sensor weight coefficient to obtain observation data of the multi-source sensor;
The error characteristic information of the system is obtained through the following steps:
determining a corresponding noise covariance matrix according to the type and the characteristics of the multi-source sensor;
Carrying out weighted average on the noise covariance matrix of each sensor according to the sensor weight coefficient to obtain a fused noise covariance matrix;
The noise covariance matrix is the error characteristic information of the system.
8. The adaptive interactive navigation positioning filtering method for multi-source sensor fusion according to claim 5, wherein the predicting and updating each sub-filtering model in the activated first filtering model or the activated second filtering model to obtain an output state of each sub-filtering model comprises:
Calculating the mixing state, the mixing state covariance matrix and the mixing model probability of each sub-filtering model according to the initial mixing probability, the initial model probability and the model transition probability matrix of each sub-filtering model;
calculating the prediction state, the prediction state covariance matrix and the prediction observation data of each sub-filtering model according to the mixed state, the mixed state covariance matrix and the system state equation;
calculating an update state, an update state covariance matrix, an update model probability and a likelihood function of each sub-filtering model according to the prediction state, the prediction state covariance matrix, the prediction observation data, the observation data and the observation equation;
And obtaining the output state and the output state covariance matrix of each sub-filtering model according to the update state, the update state covariance matrix, the update model probability and the likelihood function.
9. The adaptive interactive navigation positioning filtering method for multi-source sensor fusion according to claim 8, wherein the initial value of the input state of each sub-filtering model in the second filtering model or the first filtering model at the next moment corresponding to the preset second algorithm comprises:
calculating a residual vector of each sub-filtering model, wherein the residual vector is the difference between the observed data and the predicted observed data;
And calculating a residual covariance matrix of each sub-filtering model, wherein the residual covariance matrix is the sum of a predicted observation covariance matrix and an observation noise covariance matrix.
10. The adaptive interactive navigation positioning filtering method for multi-source sensor fusion according to claim 9, wherein the initial value of the input state of each sub-filtering model in the second filtering model or the first filtering model at the next moment corresponding to a preset second algorithm further comprises:
Calculating the average value and standard deviation of residual vectors;
Calculating the trace of the residual covariance matrix;
Determining a scaling factor based on the mean, standard deviation, and trace;
multiplying the scaling factor by the identity matrix to obtain an estimated observed noise covariance matrix;
the estimated observed noise covariance matrix is used as a new observed noise covariance matrix and is used for the filtering process at the next moment.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311644838.XA CN117906601B (en) | 2023-12-04 | 2023-12-04 | Multi-source sensor fusion-oriented adaptive interactive navigation positioning filtering method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311644838.XA CN117906601B (en) | 2023-12-04 | 2023-12-04 | Multi-source sensor fusion-oriented adaptive interactive navigation positioning filtering method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117906601A true CN117906601A (en) | 2024-04-19 |
| CN117906601B CN117906601B (en) | 2024-08-30 |
Family
ID=90695280
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311644838.XA Active CN117906601B (en) | 2023-12-04 | 2023-12-04 | Multi-source sensor fusion-oriented adaptive interactive navigation positioning filtering method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117906601B (en) |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107192995A (en) * | 2017-05-23 | 2017-09-22 | 西北工业大学 | A kind of Pure orientation underwater target tracking algorithm of multi-level information fusion |
| WO2021082571A1 (en) * | 2019-10-29 | 2021-05-06 | 苏宁云计算有限公司 | Robot tracking method, device and equipment and computer readable storage medium |
| CN114002650A (en) * | 2021-10-29 | 2022-02-01 | 哈尔滨工业大学 | Multi-sensor space-time deviation calibration and maneuvering target tracking method |
-
2023
- 2023-12-04 CN CN202311644838.XA patent/CN117906601B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107192995A (en) * | 2017-05-23 | 2017-09-22 | 西北工业大学 | A kind of Pure orientation underwater target tracking algorithm of multi-level information fusion |
| WO2021082571A1 (en) * | 2019-10-29 | 2021-05-06 | 苏宁云计算有限公司 | Robot tracking method, device and equipment and computer readable storage medium |
| CN114002650A (en) * | 2021-10-29 | 2022-02-01 | 哈尔滨工业大学 | Multi-sensor space-time deviation calibration and maneuvering target tracking method |
Non-Patent Citations (3)
| Title |
|---|
| HENK A. P. BLOM: "The Interacting Multiple Model Algorithm for Systems with Markovian Switching Coefficients", IEEE, vol. 33, no. 8, 31 August 1988 (1988-08-31) * |
| X. RONG LI: "Design of an Interacting Multiple Model Algorithm for Air Traffic Control Tracking", IEEE, vol. 1, no. 3, 30 September 1993 (1993-09-30) * |
| 赵贝贝;徐雪松;张凤云;: "基于EMA_UKF的移动机器人传感器故障诊断", 科学技术与工程, no. 05, 18 February 2017 (2017-02-18) * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117906601B (en) | 2024-08-30 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110849369B (en) | Robot tracking method, device, equipment and computer readable storage medium | |
| KR101953185B1 (en) | Multistatic passive coherent location method for estimating target location using fm network | |
| CN111127523A (en) | Multi-sensor GMPHD self-adaptive fusion method based on measurement iteration update | |
| CN108871365B (en) | State estimation method and system under course constraint | |
| Alaeiyan et al. | Hybrid noise removal to improve the accuracy of inertial sensors using lifting wavelet transform optimized by genetic algorithm | |
| CN117906601B (en) | Multi-source sensor fusion-oriented adaptive interactive navigation positioning filtering method | |
| CN106871905B (en) | Gaussian filtering substitution framework combined navigation method under non-ideal condition | |
| Maroulas et al. | A learning drift homotopy particle filter | |
| Ali et al. | Performance analysis of Gaussian optimal filtering for underwater passive target tracking | |
| CN113537302B (en) | Multi-sensor randomized data association fusion method | |
| Bhaumik et al. | Risk-sensitive formulation of unscented Kalman filter | |
| Ping et al. | Particle filter based time series prediction of daily sales of an online retailer | |
| Yuan et al. | Temporal Web Service QoS Prediction via Kalman Filter-Incorporated Latent Factor Analysis | |
| Kim et al. | Square root receding horizon information filters for nonlinear dynamic system models | |
| Morelande et al. | Smoothed state estimation for nonlinear Markovian switching systems | |
| Djuric et al. | Cost-reference particle filtering for dynamic systems with nonlinear and conditionally linear states | |
| Proletarsky et al. | Intelligent Algorithm of Measuring Complex for Aircrafts | |
| Karsaz et al. | An optimal two-stage algorithm for highly maneuvering targets tracking | |
| JP7491065B2 (en) | State estimation device, state estimation method, and state estimation program | |
| Kostoglotov et al. | Algorithm for operational processing of radar information under conditions of structural uncertainty | |
| CN116756265B (en) | Track data processing method and device, electronic equipment and storage medium | |
| CN117872269B (en) | High-precision positioning method for self-adaptive data processing | |
| CN110517297B (en) | First-order pseudo Bayesian maneuvering target tracking method based on error-fuzzy decomposition | |
| CN114719857B (en) | Gaussian aggregation filtering collaborative positioning tracking method based on constrained particle flow | |
| Li et al. | Measurement Conversion based Three-dimensional Target Tracking in Asynchronous Multi-rate Multi-radar Network |
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 |