CN116150683A - Inertial measurement unit redundancy diagnosis method based on interaction multiple models - Google Patents
Inertial measurement unit redundancy diagnosis method based on interaction multiple models Download PDFInfo
- Publication number
- CN116150683A CN116150683A CN202211542429.4A CN202211542429A CN116150683A CN 116150683 A CN116150683 A CN 116150683A CN 202211542429 A CN202211542429 A CN 202211542429A CN 116150683 A CN116150683 A CN 116150683A
- Authority
- CN
- China
- Prior art keywords
- model
- measurement unit
- inertial measurement
- probability
- 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.)
- Pending
Links
- 238000005259 measurement Methods 0.000 title claims abstract description 72
- 238000000034 method Methods 0.000 title claims abstract description 43
- 238000003745 diagnosis Methods 0.000 title claims abstract description 19
- 230000003993 interaction Effects 0.000 title claims description 8
- 230000008569 process Effects 0.000 claims abstract description 24
- 230000002452 interceptive effect Effects 0.000 claims abstract description 14
- 238000001514 detection method Methods 0.000 claims abstract description 13
- 238000004364 calculation method Methods 0.000 claims abstract description 10
- 239000011159 matrix material Substances 0.000 claims description 26
- 238000001914 filtration Methods 0.000 claims description 24
- 230000007704 transition Effects 0.000 claims description 16
- 230000004927 fusion Effects 0.000 claims description 6
- 230000001133 acceleration Effects 0.000 claims description 4
- 238000012546 transfer Methods 0.000 claims description 4
- 230000008859 change Effects 0.000 claims description 3
- 238000010606 normalization Methods 0.000 claims description 3
- 238000005070 sampling Methods 0.000 claims description 3
- 238000002405 diagnostic procedure Methods 0.000 claims 1
- 238000013461 design Methods 0.000 abstract description 5
- 238000007499 fusion processing Methods 0.000 abstract description 3
- 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 3
- 238000006243 chemical reaction Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/18—Complex mathematical operations for evaluating statistical data, e.g. average values, frequency distributions, probability functions, regression analysis
-
- 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)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Optimization (AREA)
- Pure & Applied Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Mathematical Physics (AREA)
- Computational Mathematics (AREA)
- Mathematical Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Operations Research (AREA)
- Probability & Statistics with Applications (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Algebra (AREA)
- Evolutionary Biology (AREA)
- Databases & Information Systems (AREA)
- Software Systems (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Computational Biology (AREA)
- Navigation (AREA)
Abstract
The invention relates to an inertial measurement unit redundancy diagnosis method based on an interactive multi-model, which can monitor whether the output of a master inertial measurement unit and a slave inertial measurement unit is normal or not in real time in the whole flight process. The invention innovatively provides an inertial measurement unit redundancy diagnosis method based on interactive multiple models, and solves the problems that the current online fault detection of redundant inertial measurement units usually needs to design thresholds according to different flight segments and actual trajectory and tasks, and the expansibility is poor; according to the invention, a motion sub-model is established for each flight segment of the rocket, the output results of the sub-model filters are subjected to weighted fusion processing to obtain a final estimation result, the estimation result is compared with the output of the master inertial measurement unit and the slave inertial measurement unit and the navigation calculation result, a universal threshold is used in the whole process, and whether the output of the master inertial measurement unit and the slave inertial measurement unit is normal or not can be monitored in real time in the whole flight process.
Description
Technical Field
The invention relates to the field of rocket inertial measurement unit redundancy design, in particular to an inertial measurement unit redundancy diagnosis method based on interactive multiple models.
Background
Along with technological development and technological progress, the reliability requirement of the carrier rocket is higher and higher, and the strapdown inertial measurement combination is an important single machine device of a control system and relates to success and failure of carrier rocket flight. Therefore, the adoption of the necessary redundant design for the inertial measurement unit is an important means for improving the reliability of the rocket. Currently, in the field of domestic and foreign transportation, a spacecraft mostly adopts a double-eight-meter strapdown or three-strapdown inertial measurement unit redundant control system, redundant control is realized by increasing the number of meters and then performing software and hardware diagnosis, and the defects brought by the method are the improvement of system cost and the complexity of a structure; meanwhile, the online fault detection of the redundant inertial measurement unit needs to manually design threshold parameters, and the expansibility is poor.
Disclosure of Invention
The inertial measurement unit redundancy diagnosis method based on the interactive multi-model can monitor whether the output of the master inertial measurement unit and the slave inertial measurement unit is normal or not in real time in the whole flight process. The basic principle of the interactive multi-model algorithm is that a plurality of possible motion models are established aiming at a research target, each sub-model is provided with a respective filter, the transfer process between the models is subjected to a Markov process, and the output results of the plurality of filters are subjected to weighted fusion processing to obtain a final estimation result.
The specific technical scheme of the invention is as follows: an inertial measurement unit redundancy diagnosis method based on interactive multiple models comprises the following steps:
s1, building a motion sub-model (each flight segment corresponds to a model) according to the characteristics of each flight segment of the rocket; applying an extended Kalman filtering algorithm as a sub-filtering model of each sub-model;
s2, setting model initial probability and model transition probability, using measurement data of a satellite navigation receiver for updating a measurement equation, taking rocket initial state as input of each sub-model, and calculating state estimation of each sub-filtering model;
s3, calculating updated model probability, and then carrying out weighted fusion on a plurality of state estimation values to obtain a final state estimation result, wherein covariance of each filtering model is also updated in sequence;
s4, comparing the output and navigation calculation results of the master inertial measurement unit and the slave inertial measurement unit with the state estimation results in consistency, calculating a fault detection value according to whether the difference value is larger than a set threshold value in n continuous periods, and judging whether the inertial measurement unit is normal.
Further, in the step S1,
the state equation and the measurement equation of the model i can be expressed by the following formula,
X(k+1)＝F i (k)X(k)+G i (k)ω i (k)
Z(k)＝H i (k)X(k)+v i (k)
wherein X (k+1) represents a 1×n-dimensional state vector of the object in the k+1 time model i, Z (k) represents a 1×r-dimensional observation vector of the object in the k time model i, F i (k) For the state transition matrix of model i, G i The process noise transfer matrix representing model i, H being the observed Jacobian matrix, ω i (k) Is model i process noise, v i (k) The model i measures noise, two groups of noise are independent, and k represents sampling time.
Further, the state vector X (k) includes 15 dimensions, namely, a X, Y, Z-way position, a velocity, an acceleration, an euler angle, and an angular rate of satellite navigation positioning.
Further, the observation vector Z (k) includes 6 dimensions, namely a X, Y, Z-directional position and a X, Y, Z-directional velocity of satellite navigation positioning.
Further, in the step S2, the model probability of the model i of the target at the k moment is μ i (k) Assuming that the model matched at the last moment of change of the motion rule is a model i, the next moment is a model j, and the models i to j are from model i to modelThe model transition probability of j is p ij And the conversion process between the models follows the Markov process, the probability transition matrix is:
wherein m is the number of sub-models;
the initial inputs are calculated as follows:
probability of switching from model i to model j at time k-1:
wherein ui (k-1) is the probability of the k-1 moment model i; c j Representing the prediction probability of the model j after input interaction for the normalization constant;
calculating state estimation and covariance matrix of the model j after k-1 moment input interaction:
wherein ,
representing a filtered estimate of the k-1 moment model i;
the parallel filtering of each submodel is calculated as follows:
state prediction:
wherein ,Fj (k-1) x is a state transition matrix of a k-1 moment model j, Q j Covariance matrix of process noise;
filtering gain:
wherein ,Rj Covariance matrix for measuring noise;
ε j i.e. the residual between the actual and predicted observations, S j (k) Namely, a corresponding covariance matrix;
kalman gain coefficient: k (K) j (k)＝P j (k|k-1)H j (k)S j -1 (k)
And (5) updating the state:
state filtering value:
covariance filter values: p (P) j (k|k)＝[I-K j (k)H j (k)]P j (k|k-1)
Further, in step S3, the likelihood function of the model j:
updating the probability of the model j according to a Bayesian probability formula:
outputting the fused filtering value:
outputting the covariance estimation value after fusion:
further, in step S4, when the fault detection value is greater than or equal to 6, it indicates that the inertial measurement unit is faulty, otherwise, it indicates that the inertial measurement unit is normal.
Further, in step S4, the threshold value is set according to the accuracy difference between the master inertial measurement unit and the slave inertial measurement unit.
Further, in step S4, n is 10-30, and the period is a period of navigation calculation.
Compared with the prior art, the invention has the beneficial effects that:
the invention innovatively provides an inertial measurement unit redundancy diagnosis method based on interactive multiple models, and solves the problems that the current online fault detection of redundant inertial measurement units usually needs to design thresholds according to different flight segments and actual trajectory and tasks, and the expansibility is poor; according to the invention, a motion sub-model is established for each flight segment of the rocket, the output results of the sub-model filters are subjected to weighted fusion processing to obtain a final estimation result, the estimation result is compared with the output of the master inertial measurement unit and the slave inertial measurement unit and the navigation calculation result, a universal threshold is used in the whole process, and whether the output of the master inertial measurement unit and the slave inertial measurement unit is normal or not can be monitored in real time in the whole flight process.
Drawings
FIG. 1 is a flow chart of an inertial unit redundancy diagnosis method based on an interactive multi-model;
FIG. 2 is a flowchart of an interactive multimodal algorithm.
Detailed Description
The invention will be described in detail below with reference to the drawings and the detailed description.
The invention provides an inertial unit redundancy diagnosis method based on an interactive multi-model, which has a flow shown in a figure 1 and comprises the following steps:
s1, building m motion sub-models (each flight segment corresponds to a model) according to the characteristics of each flight segment of a rocket; applying an extended Kalman filtering algorithm as a sub-filtering model of each sub-model;
the state equation and the measurement equation of the model i (i-th motion model) can be expressed by the following equation,
X(k+1)＝F i (k)X(k)+G i (k)ω i (k)
Z(k)＝H i (k)X(k)+v i (k)
wherein X (k+1) represents a 1×n-dimensional state vector of the object in the k+1 time model i, Z (k) represents a 1×r-dimensional observation vector of the object in the k time model i, F i (k) For the state transition matrix of model i, G i The process noise transfer matrix representing model i, H being the observed Jacobian matrix, ω i (k) Is model i process noise, v i (k) The model i measures noise, two groups of noise are independent, and k represents sampling time.
In this embodiment, the state vector X (k) includes 15 dimensions, i.e., n=15, where the 15 dimensions are the position, the speed, the acceleration, the euler angle, and the angular velocity of the X, Y, Z direction of the satellite navigation positioning, respectively.
In this embodiment, the observation vector Z (k) includes 6 dimensions, i.e., r=6, and the 6 dimensions are the position and the velocity of X, Y, Z directions respectively for satellite navigation positioning.
S2, as shown in FIG. 2, during the initial process, setting model initial probability (generally, the initial state is known, so the model initial probability is given empirically) and model transition probability by experience, using the measurement data of the satellite navigation receiver for updating the measurement equation, using the rocket initial state as the input of each sub-model, and calculating the state estimation of each sub-filtering model;
model probability of model i for the object at time k is mu i (k) The model matching at the last moment of the change of the motion rule is assumed to be a model i, the next moment is assumed to be a model j, and the model transition probability from the model i to the model j is given empirically to be p ij And the conversion process between the models follows the Markov process, the probability transition matrix is:
wherein m is the number of sub-models;
the initial inputs are calculated as follows:
probability of switching from model i to model j at time k-1:
wherein ,μi (k-1) is the probability of the model i at time k-1;
representing the prediction probability of the model j after input interaction for the normalization constant;
calculating state estimation and covariance matrix of the model j after k-1 moment input interaction:
wherein ,
representing a filtered estimate of the k-1 moment model i;
the parallel filtering of each submodel is calculated as follows:
(a) State prediction (intermediate state from k-1 to k):
wherein ,Fj (k-1) x is a state transition matrix of a k-1 moment model j, Q j Covariance matrix of process noise;
(b) Filtering gain:
wherein ,Rj Covariance matrix for measuring noise;
ε j i.e. the residual between the actual and predicted observations, S j (k) Namely, a corresponding covariance matrix;
kalman gain coefficient: k (K) j (k)＝P j (k|k-1)H j (k)S j -1 (k)
(c) And (5) updating the state:
state filtering value:
covariance filter values: p (P) j (k|k)＝[I-K j (k)H j (k)]P j (k|k-1)
S3, calculating updated model probability, and then carrying out weighted fusion on a plurality of state estimation values to obtain a final state estimation result, wherein covariance of each sub-filtering model is also updated in sequence;
likelihood function of model j:
updating the probability of the model j according to a Bayesian probability formula:
wherein ,
outputting the fused filtering value to obtain a final state estimation result:
outputting the covariance estimation value after fusion:
s4, comparing the output and navigation calculation results of the master inertial measurement unit and the slave inertial measurement unit with the final state estimation result in consistency, calculating a fault detection value according to whether the difference value is larger than a set threshold value in n continuous periods, and judging whether the inertial measurement unit is normal.
n periods, n being 10-30, the period being the period of navigation computation, 10 milliseconds in this embodiment; the threshold value is set according to the precision difference between the master inertial measurement unit and the slave inertial measurement unit, for example, the threshold value is in a value range when the precision difference between the master inertial measurement unit and the slave inertial measurement unit is within 1 order of magnitudeThe circumference is set to be 0.5-1km in position, 5-10m/s in speed and 0.5-1m/s in acceleration 2 Euler angle is 1-3 degrees, and angle rate is 0.5-1 degree/s; the initial value of the fault detection value is 0, the output 15-dimensional state estimation result is respectively differenced with the navigation calculation result, the fault detection value is added with 1 when the difference value is larger than the threshold value, the fault detection value is larger than or equal to 6, the inertial unit fault is represented, and otherwise, the inertial unit is normal. Taking 15 state quantities monitored in the embodiment as an example, when the deviation between 6 or more state quantities in the 15 state quantities and the navigation calculation result exceeds the threshold, the inertial measurement unit fault is determined.
When the main inertial measurement unit is normal and the auxiliary inertial measurement unit fails, performing flight control by using the main inertial measurement unit; when the master inertial measurement unit fails and the slave inertial measurement unit is normal, performing flight control by using the slave inertial measurement unit; and when the master inertial measurement unit and the slave inertial measurement unit are normal or fail, performing flight control by using the master inertial measurement unit. Steps S1, S2, S3 are repeated after the update state vector is input.
Claims (9)
1. An inertial measurement unit redundancy diagnosis method based on an interactive multi-model is characterized by comprising the following steps of:
s1, establishing a motion sub-model according to the characteristics of each flight segment of a rocket; applying an extended Kalman filtering algorithm as a sub-filtering model of each sub-model;
s2, setting initial probability and transition probability of the model, using measurement data of a satellite navigation receiver for updating a measurement equation, taking rocket initial state as input of each sub-model, and calculating state estimation of each sub-filtering algorithm;
s3, calculating updated model probability, and then carrying out weighted fusion on a plurality of state estimation values to obtain a final state estimation result, wherein covariance of each filtering model is also updated in sequence;
s4, respectively differencing the navigation calculation results of the master inertial measurement unit and the slave inertial measurement unit with the final state estimation results, calculating a fault detection value according to whether the difference value is larger than a set threshold value in n continuous periods, and judging whether the inertial measurement unit is normal.
2. The inertial measurement unit redundancy diagnosis method according to claim 1, wherein, in step S1,
the state equation and the measurement equation of the model i are expressed by the following formulas,
X(k+1)＝F i (k)X(k)+G i (k)ω i (k)
Z(k)＝H i (k)X(k)+v i (k)
wherein X (k+1) represents a 1×n-dimensional state vector of the object in the k+1 time model i, Z (k) represents a 1×r observation vector of the object in the k time model i, F i (k) For the state transition matrix of model i, G i The process noise transfer matrix representing model i, H being the observed Jacobian matrix, ω i (k) Is model i process noise, v i (k) The model i measures noise, two groups of noise are independent, and k represents sampling time.
3. An inertial redundancy diagnostic method based on an interactive multi-model according to claim 2, wherein the state vector comprises 15 dimensions, position, velocity, acceleration, euler angle and angular rate of X, Y, Z directions respectively.
4. The inertial measurement unit redundancy diagnosis method based on the interactive multi-model according to claim 2, wherein the observation vector comprises 6 dimensions, namely a X, Y, Z-directional position and a X, Y, Z-directional speed of satellite navigation positioning.
5. The inertial measurement unit redundancy diagnosis method according to claim 1, wherein in step S2, model probability of model i at time k of the target is μ i (k) The model matching at the previous moment of the change of the motion rule is assumed to be a model i, the next moment is assumed to be a model j, and p is used ij The model transition probabilities for model i to model j are represented, and the transition process between this model follows the Markov process, the probability transition matrix being represented as:
wherein m is the number of sub-models;
the initial inputs are calculated as follows:
probability of switching from model i to model j at time k-1:
wherein ,μi (k-1) is the probability of the model i at time k-1;
representing the prediction probability of the model j after input interaction for the normalization constant; />
Calculating state estimation and covariance matrix of the model j after k-1 moment input interaction:
the parallel filtering of each submodel is calculated as follows:
state prediction:
wherein ,Fj (k-1) x is a state transition matrix of a k-1 moment model j, Q j Covariance matrix of process noise;
filtering gain:
wherein ,Rj Covariance matrix for measuring noise;
kalman gain coefficient:
and (5) updating the state:
state filtering value:
covariance filter values: p (P) j (k|k)＝[I-K j (k)H j (k)]P j (k|k-1)。
6. The inertial measurement unit redundancy diagnosis method according to claim 5, wherein, in step S3,
likelihood function of model j:
updating the probability of the model j according to a Bayesian probability formula:
wherein ,
outputting the fused filtering value:
outputting the covariance estimation value after fusion:
7. the inertial unit redundancy diagnosis method based on the interactive multi-model according to claim 2, wherein in step S4, the initial value of the fault detection value is 0, the output 15-dimensional state estimation results are respectively differenced with the navigation calculation results, each time the difference is greater than the threshold value, the fault detection value is added with 1, the fault detection value is greater than or equal to 6, the inertial unit fault is represented, and otherwise, the inertial unit is represented as normal.
8. The inertial measurement unit redundancy diagnosis method according to any one of claims 1-7, wherein in step S4, the threshold value is set according to a precision difference between the master inertial measurement unit and the slave inertial measurement unit.
9. The inertial measurement unit redundancy diagnosis method according to any one of claims 1 to 7, wherein in step S4, n is 10 to 30, and the period is a period of navigation calculation.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211542429.4A CN116150683A (en) | 2022-12-02 | 2022-12-02 | Inertial measurement unit redundancy diagnosis method based on interaction multiple models |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211542429.4A CN116150683A (en) | 2022-12-02 | 2022-12-02 | Inertial measurement unit redundancy diagnosis method based on interaction multiple models |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116150683A true CN116150683A (en) | 2023-05-23 |
Family
ID=86349749
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211542429.4A Pending CN116150683A (en) | 2022-12-02 | 2022-12-02 | Inertial measurement unit redundancy diagnosis method based on interaction multiple models |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116150683A (en) |
-
2022
- 2022-12-02 CN CN202211542429.4A patent/CN116150683A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109343550B (en) | Spacecraft angular velocity estimation method based on rolling time domain estimation | |
| CN112130543B (en) | Carrier rocket electrical system based on FC-AE-1553 high-speed bus | |
| CN110288046B (en) | Fault prediction method based on wavelet neural network and hidden Markov model | |
| Chambers et al. | Robust multi-sensor fusion for micro aerial vehicle navigation in GPS-degraded/denied environments | |
| CN109472418A (en) | Maneuvering target state prediction optimization method based on Kalman filtering | |
| CN111156987A (en) | Inertia/astronomical combined navigation method based on residual compensation multi-rate CKF | |
| CN111913175A (en) | Water surface target tracking method with compensation mechanism under transient failure of sensor | |
| Yousuf et al. | Information fusion of GPS, INS and odometer sensors for improving localization accuracy of mobile robots in indoor and outdoor applications | |
| CN115857529A (en) | Actuator fault reconstruction method of spacecraft attitude control system | |
| CN110375731A (en) | A kind of mixing interacting multiple model filters method | |
| CN113190036A (en) | Unmanned aerial vehicle flight trajectory prediction method based on LSTM neural network | |
| CN112083457B (en) | Neural network optimized IMM satellite positioning navigation method | |
| CN112859898A (en) | Aircraft trajectory prediction method based on two-channel bidirectional neural network | |
| CN114923503B (en) | On-orbit spacecraft gyroscope and star sensor fault diagnosis method based on principal component analysis | |
| Han et al. | Quadratic-Kalman-filter-based sensor fault detection approach for unmanned aerial vehicles | |
| CN112833919B (en) | Management method and system for redundant inertial measurement data | |
| CN113486564B (en) | Unmanned ship propeller fault diagnosis system and method | |
| CN116993832B (en) | CNN-based position prediction method and device for flight target | |
| CN114386599A (en) | Method and device for training trajectory prediction model and trajectory planning | |
| CN109507989A (en) | A kind of method of unmanned plane sensor fault diagnosis | |
| CN116150683A (en) | Inertial measurement unit redundancy diagnosis method based on interaction multiple models | |
| CN118111430A (en) | Interactive multi-model AUV integrated navigation method based on minimum error entropy Kalman | |
| CN116576890B (en) | GNSS/INS integrated navigation system fault detection method based on integrated neural network | |
| CN116834977A (en) | Range control method for satellite orbit data | |
| CN112287560A (en) | Solver design method for rocket online trajectory planning |
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 |