CN115048371A - Fuzzy evidence track association method based on data cleaning and generation of countermeasure network - Google Patents
Fuzzy evidence track association method based on data cleaning and generation of countermeasure network Download PDFInfo
- Publication number
- CN115048371A CN115048371A CN202210978480.3A CN202210978480A CN115048371A CN 115048371 A CN115048371 A CN 115048371A CN 202210978480 A CN202210978480 A CN 202210978480A CN 115048371 A CN115048371 A CN 115048371A
- Authority
- CN
- China
- Prior art keywords
- track
- data
- association
- evidence
- fuzzy
- 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
- 238000000034 method Methods 0.000 title claims abstract description 70
- 238000004140 cleaning Methods 0.000 title claims abstract description 19
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 42
- 230000002457 bidirectional effect Effects 0.000 claims abstract description 11
- 238000012545 processing Methods 0.000 claims abstract description 11
- 238000007781 pre-processing Methods 0.000 claims abstract description 4
- 230000006870 function Effects 0.000 claims description 42
- 239000010410 layer Substances 0.000 claims description 15
- 230000008569 process Effects 0.000 claims description 12
- 230000001133 acceleration Effects 0.000 claims description 11
- 230000004913 activation Effects 0.000 claims description 9
- 239000011159 matrix material Substances 0.000 claims description 9
- 150000001875 compounds Chemical class 0.000 claims description 8
- 238000012549 training Methods 0.000 claims description 6
- 230000015572 biosynthetic process Effects 0.000 claims description 5
- 238000003786 synthesis reaction Methods 0.000 claims description 5
- 239000002356 single layer Substances 0.000 claims description 4
- 230000001364 causal effect Effects 0.000 claims description 3
- 238000013507 mapping Methods 0.000 claims description 3
- 238000012216 screening Methods 0.000 claims description 3
- 238000012546 transfer Methods 0.000 claims description 3
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 claims description 3
- 108010063499 Sigma Factor Proteins 0.000 claims description 2
- 238000005406 washing Methods 0.000 claims 3
- 230000003042 antagnostic effect Effects 0.000 claims 1
- 238000002474 experimental method Methods 0.000 description 13
- 230000010006 flight Effects 0.000 description 7
- 238000004364 calculation method Methods 0.000 description 6
- 238000010276 construction Methods 0.000 description 6
- 238000013461 design Methods 0.000 description 5
- 238000010586 diagram Methods 0.000 description 5
- 230000000694 effects Effects 0.000 description 5
- 238000012986 modification Methods 0.000 description 4
- 230000004048 modification Effects 0.000 description 4
- 230000004044 response Effects 0.000 description 4
- 230000007423 decrease Effects 0.000 description 3
- 230000008439 repair process Effects 0.000 description 3
- 230000008859 change Effects 0.000 description 2
- 238000013213 extrapolation Methods 0.000 description 2
- 238000005259 measurement Methods 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- RTAQQCXQSZGOHL-UHFFFAOYSA-N Titanium Chemical group [Ti] RTAQQCXQSZGOHL-UHFFFAOYSA-N 0.000 description 1
- 230000004075 alteration Effects 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 238000013398 bayesian method Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 230000000052 comparative effect Effects 0.000 description 1
- 238000004590 computer program Methods 0.000 description 1
- 230000002596 correlated effect Effects 0.000 description 1
- 230000000875 corresponding effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 239000000835 fiber Substances 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- YHXISWVBGDMDLQ-UHFFFAOYSA-N moclobemide Chemical compound C1=CC(Cl)=CC=C1C(=O)NCCN1CCOCC1 YHXISWVBGDMDLQ-UHFFFAOYSA-N 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/21—Design, administration or maintenance of databases
- G06F16/215—Improving data quality; Data cleansing, e.g. de-duplication, removing invalid entries or correcting typographical errors
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/29—Geographical information databases
-
- 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
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N5/00—Computing arrangements using knowledge-based models
- G06N5/04—Inference or reasoning models
- G06N5/048—Fuzzy inferencing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N7/00—Computing arrangements based on specific mathematical models
- G06N7/02—Computing arrangements based on specific mathematical models using fuzzy logic
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/40—Business processes related to the transportation industry
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G5/00—Traffic control systems for aircraft, e.g. air-traffic control [ATC]
- G08G5/0073—Surveillance aids
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Software Systems (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- Databases & Information Systems (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Molecular Biology (AREA)
- Business, Economics & Management (AREA)
- Fuzzy Systems (AREA)
- Automation & Control Theory (AREA)
- Biomedical Technology (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Economics (AREA)
- Computational Mathematics (AREA)
- Mathematical Analysis (AREA)
- Mathematical Optimization (AREA)
- Pure & Applied Mathematics (AREA)
- Aviation & Aerospace Engineering (AREA)
- Quality & Reliability (AREA)
- Remote Sensing (AREA)
- Algebra (AREA)
- Human Resources & Organizations (AREA)
- Marketing (AREA)
- Primary Health Care (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
The invention relates to the technical field of flight path processing, in particular to a fuzzy evidence flight path correlation method based on data cleaning and generation of a countermeasure network. The method comprises the following steps: if the current track is interrupted, preprocessing the suspicious track and track data before interruption when the suspicious track appears again, and inputting the preprocessed data into a pre-constructed data cleaner for data cleaning; inputting the track data output by the data cleaner into a pre-constructed countermeasure network model, performing bidirectional track prediction on the track data, and outputting the bidirectional predicted track data; and performing track association on the track data before interruption and the bidirectionally predicted track data by using a fuzzy evidence interruption track association algorithm. The track data discontinuity problem in track data processing that the track label attribute changes and the track breaking period is long after the track breaking can be effectively associated.
Description
Technical Field
The invention relates to the technical field of flight path processing, in particular to a fuzzy evidence flight path correlation method based on data cleaning and generation of a countermeasure network.
Background
The flight path of the airplane generally has the characteristics of high density, high speed, low relative speed among targets, poor separability and the like, and the phenomenon of flight path data interruption can occur under the influence of uncertain factors such as sensor characteristics, geographical environment, electromagnetic interference, electromagnetic silence, information countermeasure and the like. After the attribute of the target label of the interrupted track is changed, the track data seriously affects the information fusion; and the interruption of the flight path enables the sensor to compile and track the target again, thus increasing the tracking load of the sensor and causing the low efficiency of tracking and measurement. The method solves the problem of track association before and after the same target is interrupted, can improve the target tracking continuity and stability, and can realize real-time tracking of the target.
The association problem of broken track was proposed in the 80 s of the 20 th century, and the track association under different conditions in recent years is the focus of research of scholars at home and abroad. The following problems exist in different practical situations of the interrupted track association algorithm: firstly, due to the existence of system and measurement errors, the backward prediction error of a new track is often large, so that the accuracy of association pairing of the new track and an old track based on a single point is poor, and the occurrence of miscorrelation and missed association is frequent; secondly, when the targets are dense, the targets are influenced by factors such as track intersection and bifurcation, so that the cross correlation often occurs, and the performance of the algorithm for correlating the interrupted track suddenly decreases as the track interruption time increases.
The traditional algorithm has the characteristics of large calculation amount and high requirement on target maneuvering prior information, is not suitable for a long-term tracking environment with concentrated targets, and has poor robustness. With the increasingly complex tracking environment, the algorithm is inevitably developed towards the direction of appropriate calculated amount and low requirement on target prior information. Therefore, the problem of track data discontinuity in track data processing that the track label attribute changes after track breaking and the track breaking period is long is a problem which needs to be solved urgently in the field of track processing.
Disclosure of Invention
Aiming at the defects of the prior art, the invention provides a fuzzy evidence track association method based on data cleaning and generation of a countermeasure network, which comprises the steps of firstly, performing data cleaning on track data; secondly, inputting the characteristic vector into an improved generation countermeasure network model to complete prediction and bidirectional extrapolation of track data, and finally performing track segment association by adopting a fuzzy evidence track association algorithm, so that track segments before and after track breakage can be effectively associated, and the problem of track data discontinuity in track data processing of track label attribute change and track breakage period after track breakage is solved.
The invention discloses a fuzzy evidence track association method based on data cleaning and generation of a countermeasure network, which adopts the technical scheme that: the method comprises the following steps:
if the current track is interrupted, preprocessing the suspicious track and track data before interruption when the suspicious track appears again, and inputting the preprocessed data into a pre-constructed data cleaner for data cleaning;
inputting the flight path data output by the data cleaner into a pre-constructed countermeasure network model, performing bidirectional flight path prediction on the flight path data, and outputting the bidirectionally predicted flight path data;
performing track association on track data before interruption and track data predicted bidirectionally by using a fuzzy evidence interruption track association algorithm;
if the association is successful, ending the process;
if the association fails, updating the flight path association algorithm parameters, performing secondary association on the flight path, ending the process until the association succeeds, or ending the process when the number of times of repeatedly performing secondary association reaches the maximum re-association number.
Preferably, the method for pre-constructing the data washer includes:
selecting a set of random variables characterizing a problemX 1 ,X 2 ,…,X n And selecting the sequence;
starting from an empty figure, adding the variables one by one according to the selected variable sequenceξPerforming the following steps;
selecting a minimum subset pi (in) in the variable set by using the prior information of the problemX n ) So that "given π (X n ) When the temperature of the water is higher than the set temperature,X n andξthe other variables in (1) are independent of the condition;
from pi (X n ) Adding a point to each node inX n Obtaining a directed acyclic graph model by the directed edges;
and performing parameter learning on the directed acyclic graph model to obtain a conditional probability distribution set of each variable.
Preferably, the priori information is obtained by encoding, and the encoding process includes:
searching a causal relationship among all characteristics in the data;
defining each feature and its associated condition as a class;
inquiring Query to indicate how to get the class to which the Query belongs and the attribute to which the class belongs from the value of each item in the data;
obtaining a class relation network of prior information;
and further screening the result which is obtained by the Bayesian network and does not accord with the logic based on the encoding result of the prior information by the class relation network.
Preferably, the method for constructing the confrontation network model in advance includes:
aircraft to be cleaned by data cleaneriTraining data of a flight path sequenceInputting into a single-layer MLP with an activation function;
translating a single dimension of an aircraft into a fixed length vectorObtaining attention weight in each dimension using attention moduleApplied to the input to obtain a spatial vector of interest;
Will focus on the space vectorMapping into potential space through embedded layer to obtain spatial feature vector capable of transmitting to generate countermeasure networkThe long-short term memory network LSTM passing the vector and the last stateCoding to obtain hidden features of the aircraftFinally, embedding the signal into a sample space through MLP (Multi-level Linear Power) and superposing random noise b to obtain the input of a generator;
Calculating relative position information of the aircraft, and then coding the position information by using a convergence layer to obtain a convergence matrixA i In the prediction oftAt the time of aircraft position, first willtPredicted position at time-1Converting the embedded function into a feature space to obtain featuresWill beA i Andt-hidden feature of aircraft at time 1Using the MLP coding result as the hidden feature of the decoding layer, and comparing the hidden feature with the feature of the decoding layerAre input into a decoding layer together to obtainFinally, obtaining the predicted flight path through MLP function coding;
Coding the result of the generator to obtain hidden features, decoding the hidden features through an MLP function, and calculating an output result by using a Softmax classifier to obtain an output score;
the loss is calculated by a loss function and the model is trained by inverse transfer gradient tuning parameters.
Preferably, the performing, by using the fuzzy evidence interrupt track association algorithm, track association between the track data before the interrupt and the two-way predicted track data includes:
determining a fuzzy factor set;
performing track association by using a normal type membership function, and adjusting the spread of position, speed and acceleration factors;
calculating the correlation degree of a certain flight path before interruption and a certain flight path after interruptionf ij (l) And based on a first targetnOf a track and a second targetmEach track, constituting a fuzzy correlation matrix at a time:
If the largest elementf ij (l) Greater than a set relevancy thresholdεThen flight pathiAndjthe association is successful; otherwise, the track is determinediAndjthe association fails.
Preferably, after the track association is performed on the track data before the interruption and the track data of the bidirectional prediction by using the fuzzy evidence interruption track association algorithm to obtain a preliminary result, the result is corrected by combining an evidence theory.
Preferably, the correcting the result by the combined evidence theory includes:
setting the identification frame as aθ 1 ,θ 2 ,θ 3 },θ 1 ,θ 2 Andθ 3 respectively expressed as association, non-association and uncertainty, and three evidence bodies are constructed according to the position, the speed and the acceleration, namely:
calculating similarity between evidence bodiesc(m i ,m j );
Calculating respective evidence bodiesm i Supported degree of (c) ((c))m i );
For each evidence bodym i The supported degree is normalized to obtain evidence discountw(m i );
Evidence-based discountsw(m i ) Carrying out weighted synthesis on the evidence;
according to D-S evidence theory on evidence bodymTo carry outn1 synthesis to obtain the final combined resultm(θ i ),i(= 1, 2, 3) whenm(θ 1 )- m(θ 2 )> m(θ 3 ) When considering the sample association。
Preferably, the updating the flight path association algorithm parameter and the performing the secondary association of the flight path includes:
adding newly received flight path data, and further expanding the flight path data after processing;
membership function to second fuzzy factorμ 2 Updating to make it to the interrupt timetIs more sensitive, the membership functionμ 2 Expressed as:
in the formula (I), the compound is shown in the specification,ν (t) Representing the time influencing factor in the second blurring factor,σ x 、σ y 、σ z the spread of the position ambiguity factor is represented,τ 2 in order to adjust the degree of the adjustment,u 2 is a blurring factor representing speed;
according to interrupt time intervaltIs set with a correlation thresholdεThe relevance threshold is expressed as:ε=1- f(t);
after the track association algorithm parameters are updated, the track association algorithm is interrupted by the fuzzy evidence again to carry out track association on the track data before interruption and the two-way predicted track data.
The invention has the beneficial effects that:
1. carrying out data cleaning on the track data; secondly, inputting the characteristic vector into an improved generation countermeasure network model to complete prediction and bidirectional extrapolation of track data, and finally performing track segment association by adopting a fuzzy evidence track association algorithm, so that track segments before and after track breakage can be effectively associated, and the problem of track data discontinuity in track data processing of track label attribute change and track breakage period after track breakage is solved.
2. In the aspect of a data cleaner, the method combines the Bayesian network and the coding of the prior knowledge, and compared with the traditional method, the prior knowledge can be introduced less and more conveniently, so that the usability and the portability of the method are enhanced. In addition, the data cleaner can obtain the result which is more in accordance with the actual situation and can not obtain the result which deviates from the actual situation. In addition, the probability of error cleaning of the correct data by the model is reduced by encoding the prior knowledge. The program time is significantly reduced compared to conventional data cleansing methods.
3. In the aspect of constructing a generation countermeasure network model aiming at flight path prediction, a loss function of the flight path prediction is introduced into a loss function of the method, so that the generation countermeasure network can be applied to the field of flight path prediction; in addition, the loss of the generation countermeasure network and the track prediction loss are balanced by introducing two proportionality coefficients. Compared with the traditional generation of a countermeasure network model, the model of the method has the capability of flight path prediction, and in addition, due to the addition of the proportionality coefficient, the model is not biased to one type of capability, so that the model with better prediction accuracy is obtained.
4. In the aspect of a track prediction method, the method designs a bidirectional track prediction method based on generation of a countermeasure network, the method can predict the track from two ends of an interrupted track, and particularly, in the aspect of reverse prediction, the method can perform a large number of track point prediction tasks by using only a few track points. Compared with other methods, the method has better usability and faster track prediction speed.
5. In the aspect of track association, the method combines fuzzy association and evidence theory, so that higher association accuracy can be obtained. In addition, on the basis of the fuzzy association method, a multiple association mechanism is introduced on the basis of the original fuzzy association, and when the first round of track prediction fails, the probability of successful association is increased by introducing more track data, reducing an association threshold value and the like, so that the accuracy of track association is improved.
Drawings
FIG. 1 is a schematic flow diagram of the present invention;
FIG. 2 is a schematic flow chart of the track association of the present invention;
FIG. 3 is a diagram of raw data before being processed by the Bayesian data cleaning system of the present invention
FIG. 4 is a graph of repair data processed by the Bayesian data cleaning system of the present invention;
FIG. 5 is a graph illustrating the effect of different flight numbers on accuracy;
FIG. 6 is a diagram illustrating the influence of different prediction points on accuracy;
FIG. 7 is a diagram illustrating the effect of different interrupt times on the correlation accuracy.
Detailed Description
In order to make the technical problems, technical solutions and advantageous effects to be solved by the present application clearer, the present application is further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the present application and are not intended to limit the present application.
The terms "first", "second" and "first" are used for descriptive purposes only and are not to be construed as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of the present application, "a plurality" means two or more unless specifically limited otherwise.
In the following description, for purposes of explanation and not limitation, specific details are set forth, such as particular system structures, techniques, etc. in order to provide a thorough understanding of the embodiments of the present application. It will be apparent, however, to one skilled in the art that the present application may be practiced in other embodiments that depart from these specific details. In other instances, detailed descriptions of well-known systems, devices, circuits, and methods are omitted so as not to obscure the description of the present application with unnecessary detail.
It will be understood that the terms "comprises" and/or "comprising," when used in this specification and the appended claims, specify the presence of stated features, integers, steps, operations, elements, and/or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and/or groups thereof.
It should also be understood that the term "and/or" as used in this specification and the appended claims refers to and includes any and all possible combinations of one or more of the associated listed items.
As used in this specification and the appended claims, the term "if" may be interpreted contextually as "when", "upon" or "in response to" determining "or" in response to detecting ". Similarly, the phrase "if it is determined" or "if a [ described condition or event ] is detected" may be interpreted contextually to mean "upon determining" or "in response to determining" or "upon detecting [ described condition or event ]" or "in response to detecting [ described condition or event ]".
Furthermore, in the description of the present application and the appended claims, the terms "first," "second," "third," and the like are used for distinguishing between descriptions and not necessarily for describing or implying relative importance.
Reference throughout this specification to "one embodiment" or "some embodiments," or the like, means that a particular feature, structure, or characteristic described in connection with the embodiment is included in one or more embodiments of the present application. Thus, appearances of the phrases "in one embodiment," "in some embodiments," "in other embodiments," or the like, in various places throughout this specification are not necessarily all referring to the same embodiment, but rather "one or more but not all embodiments" unless specifically stated otherwise. The terms "comprising," "including," "having," and variations thereof mean "including, but not limited to," unless expressly specified otherwise. "plurality" means "two or more".
Before the method is executed, a data cleaner and a countermeasure network model need to be constructed in advance.
The process of constructing the data washer comprises the following steps:
firstly, a data cleaner based on a Bayesian network is constructed based on existing training data, and comprises structure construction and parameter learning, wherein a graph model, namely a directed acyclic graph, is determined through the structure construction, and a probability parameter table is determined through the parameter learning. In addition, the data washer also supports the user to code the prior knowledge, and the model codes the background knowledge of the data and the possible problems by the user. The scrubber may be given background knowledge about possible corruption of data and further screened for non-logical results.
The data cleaner comprises two parts of construction of a Bayesian network and encoding of a priori knowledge.
The construction of the Bayesian network comprises structure construction and parameter learning, wherein the structure construction comprises the following steps:
the method comprises the following steps: selecting a set of random variables characterizing a problemX 1 ,X 2 ,…,X n And the order is selected.
Step two: starting from an empty figure, adding the variables one by one according to the selected variable sequenceξIn (1).
Step three: selecting a minimum subset pi (in) in the variable set by using the prior information of the problemX n ) So that "given π (X n ) When the temperature of the water is higher than the set temperature,X n andξthe other variables in (1) are independent of the condition.
Step four: from pi (X n ) Adding a point to each node inX n So as to construct a directed acyclic graph model.
Step five: and performing parameter learning on the directed acyclic graph model to obtain a conditional probability distribution set of each variable. The parameter learning of the bayesian network can be generally divided into two categories, namely a likelihood estimation method and a bayesian method. The likelihood estimation method is the most widely applied method, and the present embodiment performs parameter learning using the likelihood estimation method.
In order to further screen the cleaner for non-logical results with respect to the background knowledge that the data may be corrupted, the present solution further encodes the prior knowledge, which includes the following steps:
the method comprises the following steps: and manually searching the causal relationship among the characteristics in the data.
Step two: each feature and its association is specified as a class.
Step three: the Query (Query) indicates how to derive which attribute in which class from the value of each item in the data.
Step four: and obtaining a class relation network of prior information.
Step five: and based on the encoding result of the prior information in the step four, further screening the result which is obtained by the Bayesian network and does not accord with the logic.
The method for constructing the confrontation network model in advance comprises the following steps:
the method comprises the following steps: aircraft to be cleaned by data cleaneriTraining data of a flight path sequenceInput into a single layer MLP with an activation function.
Step two: translating a single dimension of an aircraft into a fixed length vectorObtaining attention weight in each dimension using attention moduleApplied to the input to obtain a spatial vector of interest. The correlation formula is:
in the formula (I), the compound is shown in the specification,ψ(.) is a single-layer MLP embedding function with an activation function;is thatψ(.) operating parameters at longitude x;are respectively provided withIs composed ofψ(.) operation parameters in dimension and height;are respectively aircraftiIn thattCalculating weight values of attention of the time on longitude x, latitude y and height z;φ(.) is a Softmax function.
Step three: will focus on the space vectorMapping into potential space through embedded layer to obtain spatial feature vector capable of transmitting to generate countermeasure networkThe LSTM passing the vector and the last stateAnd (6) coding is carried out. Obtaining hidden features of an aircraftFinally, embedding the signal into a sample space through MLP (Multi-level Linear Power) and superposing random noise b to obtain the input of a generator. The correlation formula is:
where phi () is an embedding function with an activation function,W ee are network parameters embedded in the computation.W encoder Are the parameters of the LSTM encoding calculation,ψ(.) is an MLP embedding function with an activation function, W c is a network parameter at the time of MLP embedding computation.
Step four: calculating relative position information of the aircraft, and then coding the position information by using a convergence layer to obtain a convergence matrixA i . In the predictiontAt the time of aircraft position, first willtPredicted position at time-1(ii) a Converting the embedded function into a feature space to obtain features. Will be provided withA i Andthidden features of the aircraft at time-1Using the MLP coding result as the hidden feature of the decoding layer, and comparing the hidden feature with the feature of the decoding layerAre input into a decoding layer together to obtainFinally, obtaining the predicted flight path through MLP function coding. The correlation formula is:
where phi () is an embedding function with an activation function,W ed is the network parameter embedded in the computation and,ψ(.) is an MLP embedding function with an activation function,W decoder are the parameters of the LSTM encoding calculation,W c is a network parameter at the time of MLP embedding computation.
Step five: and coding the result of the generator to obtain hidden features, decoding the hidden features through an MLP function, and calculating an output result by using a Softmax classifier to obtain an output score. The correlation formula is:
in the formula (I), the compound is shown in the specification,W encoder are the parameters of the LSTM encoding calculation,W fc is a FC functionϕ(.) is processed in a single-pass operation,W MLP is a classifier with an activation function MLPψ(.).
Step six: the loss is calculated by a loss function and the parametric training model is adjusted by passing the gradient backwards. The correlation formula is:
wherein G, D represents the generator and the discriminator,、representing the loss functions of the generator and the arbiter, respectively.
The training steps of the backward prediction model are the same as above.
After both the data washer and the countermeasure network model are built, the method begins. The program receives the flight path data sent by the aircraft at intervals, and when the flight path data is not received within the set time, namely the current flight path is interrupted, the process of the method is started to be executed.
Example one
Fig. 1 is a schematic structural diagram illustrating a fuzzy evidence track association method based on data cleansing and generation of an countermeasure network according to a preferred embodiment of the present application. For convenience of explanation, only the parts related to the present embodiment are shown, and detailed as follows:
the specific implementation of the algorithm is as follows.
Step 1: if the current track is interrupted, when the suspicious track appears again, preprocessing the suspicious track and the track data before interruption, and inputting the preprocessed data into a pre-constructed data cleaner for data cleaning.
Step 2: and inputting the flight path data output by the data cleaner into a pre-constructed countermeasure network model, performing bidirectional flight path prediction on the flight path data, and outputting the bidirectionally predicted flight path data.
And step 3: performing track association on track data before interruption and track data predicted bidirectionally by using a fuzzy evidence interruption track association algorithm; if the association is successful, exiting the algorithm; if the association fails, step 4 is executed.
And 4, step 4: updating the flight path association algorithm parameters, and performing secondary association on the flight path; if the association still fails, firstly judging whether the number of times of the association is larger than the maximum number of times of the re-association, if not, repeating the step 4, and if the maximum number of times is reached, exiting the algorithm and failing to associate.
In one embodiment, as shown in fig. 2, step 3 mainly includes two parts, namely a fuzzy track algorithm and an evidence theory, where the fuzzy track algorithm is specifically as follows:
step 301: determining fuzzy factor set, wherein fuzzy factors u1, u2 and u3 respectively represent fuzzy factors of position, speed and acceleration
Step 302: and performing track association by using a normal type membership function, adjusting the spread of the position, speed and acceleration factors, wherein the membership functions of the position, speed and acceleration factors are respectively expressed as:
in the formula (I), the compound is shown in the specification,σ x 、σ y 、σ z representing the spread of the position ambiguity factor, i.e. the position error variance;、、representing the spread of the velocity ambiguity factor, i.e. the velocity error variance;、、representing the spread of the acceleration fuzzy factor, namely the acceleration error variance;τ 1 、τ 2 、τ 3 indicating the degree of adjustment;
step 303: calculating a certain track before interruptioniWith a certain track after interruptionjIs expressed as:
in the formula (I), the compound is shown in the specification,f ij (l) Is shown aslAt the moment of time, the time of day,iandjthe degree of correlation between two flight paths is determined bylTime of daykDegree of membershipμ 1 、μ 2 、μ 3 And corresponding weightsa 1 、a 2 、a 3 The sum of the products; for object 1 and object 2nA track andma track formed bylFuzzy correlation matrix of time:
step 304: finding the largest element in the fuzzy association matrixf ij (l) For a set threshold of degree of associationεIf, iff ij (l)>εThen representsiAndjthe two targets are associated, and the association is successful; otherwise, the association is not associated, and the association fails.
After a preliminary result is obtained by interrupting track association, the result needs to be corrected by combining an evidence theory, and the method comprises the following specific steps:
step 305: setting the identification frame as aθ 1 ,θ 2 ,θ 3 },θ 1 ,θ 2 Andθ 3 expressed as associated, unassociated and uncertain respectively, 3 evidence bodies can be constructed from position, velocity, acceleration, namely:
Step 308: normalization processing is carried out on the supported degree to obtain evidence discountw(m i );
step 310: the evidence body is synthesized for the second time according to the D-S evidence theory to obtain the final combination resultm(θ i ),i1, 2, 3, only ifm(θ 1 )- m(θ 2 )> m(θ 3 ) Samples are considered to be correlated.
And starting a secondary association algorithm when the association of the track association algorithm fails, wherein the specific algorithm steps are as follows:
step 401: and adding newly received flight path data, and further expanding the flight path data after performing relevant processing.
Step 402: updating the membership function to be the membership function for the 2 nd fuzzy factorμ 2 Adjusted to the interrupt timetIs relatively sensitive to changes in; function of degree of membershipμ 2 Expressed as:
in the formula (I), the compound is shown in the specification,ν (t) Representing the time-influencing factor in the 2 nd ambiguity factor, and the interruption time intervaltIn direct proportion, i.e. the longer the interruption time,ν (t) The greater the degree of influence of (c).
Step 403: according to interrupt time intervaltSets a threshold value, which is expressed as:
ε=1- f (t)
in the formula (I), the compound is shown in the specification,f (t) And interval of interruptiontProportional, the longer the flight path break time, the thresholdεThe smaller.
Step 404: after the relevant parameters are updated, the interrupted track association algorithm is repeatedly executed. And judging whether to start the secondary association algorithm again according to the result.
Step 405: if the re-association fails, firstly judging whether the re-association times are exceeded or not, if not, calling the secondary association algorithm again to step 401, and if so, exiting the algorithm.
Example two
This example is used to perform experimental verification on the method of the first example.
In the embodiment, the data set adopts a real airplane track data set to verify, train and generate a model for correlation between the confrontation network prediction and the fuzzy evidence, the airplane track data is composed of a plurality of data sources, one hour of all flight navigation data of a certain day around the world is extracted as a data source, flights with the flight track data length larger than 50 minutes are screened out, and 1259 pieces of flight data are finally obtained. Each piece of data contains time of flight, altitude, speed, heading, longitude, latitude information. The data set has high authenticity and reliability.
This embodiment assumes in the experiment that all flights lose signal at 200 th flight data and recover signal at 300 th flight data. The sending interval of each navigation data is about 10 seconds, namely, the flight path is interrupted for about 16.7 minutes. Each piece of navigation data is processed into time, position information in three spatial directions, speed information in three spatial directions and acceleration information in three spatial directions. And all data were normalized.
The CPU of the computer is i7-7800X, and the GPU is TITAN XP. And the track association experiment evaluates and displays the performance of the model by using three indexes of different flight numbers, different prediction points and different interruption time. The following are the settings for the specific parameters of each experiment.
Data cleaning experimental data of this example is from opensky-network org, which is a data source for the following algorithms of this example and will be used below to verify the performance of our algorithm. To demonstrate the performance of the algorithm, we artificially added some errors to verify its performance. The specific modification is shown in a result graph. We mainly modified the data icao24, squark, callsign. Constraint relationships are written as a priori information by the relationship between these features.
Some other operations may be required before the track data is flushed using the data washer. The method comprises the steps of dividing original data into different files according to flight numbers, discarding some empty data rows which cannot be repaired, normalizing the data and the like. Finally, the results of the Bayesian data cleansing system are compared with the original data and repair data pairs of the graph as shown in FIGS. 3 and 4.
The comparative prediction models of the track correlation experiment in this embodiment are TCN, BiLSTM, ARIMA, and Prophet, respectively, and Baseline performs track correlation by using only fuzzy evidence correlation without using a prediction method.
As shown in fig. 5, in the experiment of the influence of different flight numbers on the correlation accuracy, the interruption time was about 9 minutes at the 100 th data point, and 150 data points were recovered. 30 data points are predicted when the flight path is interrupted. The parameters of the prediction model are the same as those in the previous prediction experiments.
As shown in fig. 6, in the experiment of the influence of different prediction points on the correlation accuracy, 130 data points are recovered after the 100 th data point is interrupted, and the interruption time is about 6 minutes. The number of flights is 20 flights randomly selected. The parameters of the prediction model are the same as those in the previous prediction experiments.
As shown in fig. 7, in the experiment of the influence of different interruption times on the correlation accuracy, 30 points are predicted when the flight path is interrupted, and the number of flights is 30 flights randomly selected. The parameters of the prediction model were the same as those in the preceding prediction experiments.
The results of the experiment were analyzed as follows:
from fig. 3 and 4, it can be seen that the effect of the bayesian-based data cleaning system on the track data is achieved, and it can be seen that the icao24, squark and callsign feature columns of the original data contain many wrong data, and a good effect is achieved after the data are repaired. The repair accuracy rate reaches 99.85%, and the expected design requirement is met.
As can be seen in FIG. 5, after 100 data points are interrupted (about 16.7 minutes), the accuracy associated with 10 lanes simultaneously reaches about 86%, while the accuracy associated with 100 lanes simultaneously reaches about 63%. It is worth mentioning that since the flight data volume is huge, calculation using all screened routes brings huge calculation amount, so in order to develop comparison experiments, a certain number of routes are randomly selected for experiments. And if the algorithm were to predict too much information about the discontinuity in the test, this could result in a reduction in its associated accuracy due to a wrong prediction.
As can be seen from fig. 6, the algorithm has a high correlation accuracy in the case of interrupting 30 data points (about 5 minutes) while correlating 20 flights, and the accuracy increases with the increase of the prediction points, and it is worth mentioning that in practical applications, too many points should not be predicted, which may result in a decrease of the correlation success rate. In sum, the correlation algorithm has better capability of resisting the influence caused by various interrupt times.
As can be seen in FIG. 7, the 10 minute interrupt accuracy reaches 95% with 50 predicted points per lane associated with 100 lanes simultaneously. As the interruption time increases, its accuracy may show a predictable trend to decline due to its increased uncertainty. Incidentally, the program has more hyper-parameters, and the determination of the hyper-parameters is directly related to the accuracy of the program. A good selection of hyper-parameters may lead to better results.
It should be understood that the specific order or hierarchy of steps in the processes disclosed is an example of exemplary approaches. Based upon design preferences, it is understood that the specific order or hierarchy of steps in the processes may be rearranged without departing from the scope of the present disclosure. The accompanying method claims present elements of the various steps in a sample order, and are not intended to be limited to the specific order or hierarchy presented.
In the foregoing detailed description, various features are grouped together in a single embodiment for the purpose of streamlining the disclosure. This method of disclosure is not to be interpreted as reflecting an intention that the claimed embodiments of the subject matter require more features than are expressly recited in each claim. Rather, as the following claims reflect, invention lies in less than all features of a single disclosed embodiment. Thus, the following claims are hereby expressly incorporated into the detailed description, with each claim standing on its own as a separate preferred embodiment of the invention.
The previous description of the disclosed embodiments is provided to enable any person skilled in the art to make or use the present invention. To those skilled in the art; various modifications to these embodiments will be readily apparent, and the generic principles defined herein may be applied to other embodiments without departing from the spirit or scope of the disclosure. Thus, the present disclosure is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
What has been described above includes examples of one or more embodiments. It is, of course, not possible to describe every conceivable combination of components or methodologies for purposes of describing the aforementioned embodiments, but one of ordinary skill in the art may recognize that many further combinations and permutations of various embodiments are possible. Accordingly, the embodiments described herein are intended to embrace all such alterations, modifications and variations that fall within the scope of the appended claims. Furthermore, to the extent that the term "includes" is used in either the detailed description or the claims, such term is intended to be inclusive in a manner similar to the term "comprising" as "comprising" is interpreted when employed as a transitional word in a claim. Furthermore, any use of the term "or" in the specification of the claims is intended to mean a "non-exclusive or".
Those of skill in the art will further appreciate that the various illustrative logical blocks, units, and steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware, computer software, or combinations of both. To clearly illustrate the interchangeability of hardware and software, various illustrative components, elements, and steps have been described above generally in terms of their functionality. Whether such functionality is implemented as hardware or software depends upon the particular application and design requirements of the overall system. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present embodiments.
The various illustrative logical blocks, or elements, described in connection with the embodiments disclosed herein may be implemented or performed with a general purpose processor, a digital signal processor, an Application Specific Integrated Circuit (ASIC), a field programmable gate array or other programmable logic device, discrete gate or transistor logic, discrete hardware components, or any combination thereof designed to perform the functions described herein. A general-purpose processor may be a microprocessor, but in the alternative, the processor may be any conventional processor, controller, microcontroller, or state machine. A processor may also be implemented as a combination of computing devices, e.g., a digital signal processor and a microprocessor, a plurality of microprocessors, one or more microprocessors in conjunction with a digital signal processor core, or any other similar configuration.
The steps of a method or algorithm described in connection with the embodiments disclosed herein may be embodied directly in hardware, in a software module executed by a processor, or in a combination of the two. A software module may be stored in RAM memory, flash memory, ROM memory, EPROM memory, EEPROM memory, registers, hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art. For example, a storage medium may be coupled to the processor such that the processor can read information from, and write information to, the storage medium. In the alternative, the storage medium may be integral to the processor. The processor and the storage medium may reside in an ASIC, which may be located in a user terminal. In the alternative, the processor and the storage medium may reside in different components in a user terminal.
In one or more exemplary designs, the functions described above in connection with the embodiments of the invention may be implemented in hardware, software, firmware, or any combination of the three. If implemented in software, the functions may be stored on or transmitted over as one or more instructions or code on a computer-readable medium. Computer-readable media includes both computer storage media and communication media that facilitate transfer of a computer program from one place to another. Storage media may be any available media that can be accessed by a general purpose or special purpose computer. For example, such computer-readable media can include, but is not limited to, RAM, ROM, EEPROM, CD-ROM or other optical disk storage, magnetic disk storage or other magnetic storage devices, or any other medium which can be used to carry or store program code in the form of instructions or data structures and which can be read by a general-purpose or special-purpose computer, or a general-purpose or special-purpose processor. Additionally, any connection is properly termed a computer-readable medium, and, thus, is included if the software is transmitted from a website, server, or other remote source via a coaxial cable, fiber optic cable, twisted pair, Digital Subscriber Line (DSL), or wirelessly, e.g., infrared, radio, and microwave. Such discs (disk) and disks (disc) include compact disks, laser disks, optical disks, DVDs, floppy disks and blu-ray disks where disks usually reproduce data magnetically, while disks usually reproduce data optically with lasers. Combinations of the above may also be included in the computer-readable medium.
The above-mentioned embodiments are only used to illustrate the technical solutions of the present application, and not to limit the same; although the present application has been described in detail with reference to the foregoing embodiments, it should be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; such modifications and substitutions do not substantially depart from the spirit and scope of the embodiments of the present application and are intended to be included within the scope of the present application.
Claims (8)
1. A fuzzy evidence track association method based on data cleaning and generation of a countermeasure network is characterized by comprising the following steps:
if the current track is interrupted, preprocessing the suspicious track and track data before interruption when the suspicious track appears again, and inputting the preprocessed data into a pre-constructed data cleaner for data cleaning;
inputting the track data output by the data cleaner into a pre-constructed countermeasure network model, performing bidirectional track prediction on the track data, and outputting the bidirectional predicted track data;
performing track association on track data before interruption and track data predicted bidirectionally by using a fuzzy evidence interruption track association algorithm;
if the association is successful, ending the process;
if the association fails, updating the flight path association algorithm parameters, performing secondary association on the flight path, ending the process until the association succeeds, or ending the process when the number of times of repeatedly performing secondary association reaches the maximum number of times of re-association.
2. The method for associating data-based cleansing with generating fuzzy evidence tracks for antagonistic networks according to claim 1, wherein the method for pre-building the data washer comprises:
selecting a set of random variables characterizing a problemX 1 ,X 2 ,…,X n And selecting the sequence;
starting from an empty figure, adding the variables one by one according to the selected variable sequenceξPerforming the following steps;
selecting a minimum subset pi (in) in the variable set by using the prior information of the problemX n ) So that "given π (X n ) When the temperature of the water is higher than the set temperature,X n andξthe other variables in (1) are independent of the condition;
from pi (X n ) Adding a point to each node inX n Obtaining a directed acyclic graph model by the directed edges;
and performing parameter learning on the directed acyclic graph model to obtain a conditional probability distribution set of each variable.
3. The method for associating fuzzy evidence track based on data washing and generation countermeasure network as claimed in claim 2, wherein the prior information is obtained by encoding, the encoding process comprises:
searching a causal relationship among all characteristics in the data;
defining each feature and its associated condition as a class;
inquiring Query to indicate how to get the class to which the Query belongs and the attribute to which the class belongs from the value of each item in the data;
obtaining a class relation network of prior information;
and further screening the results which are not in accordance with the logic and obtained by the Bayesian network based on the encoding results of the prior information by the class relationship network.
4. The method for associating fuzzy evidence track based data cleansing and generation of confrontational network as claimed in claim 1, wherein the method for pre-constructing the confrontational network model comprises:
aircraft to be cleaned by data cleaneriTraining data of a flight path sequenceInputting into a single-layer MLP with an activation function;
translating a single dimension of an aircraft into a fixed length vectorObtaining attention weight in each dimension using attention moduleApplied to the input to obtain a spatial vector of interest;
Will focus on the space vectorMapping the embedded layer into a potential space to obtain a spatial feature vector which can be transmitted to generate a countermeasure networkThe long-short term memory network LSTM passing the vector and the last stateCoding to obtain hidden features of the aircraftFinally, embedding the signal into a sample space through MLP (Multi-level Linear Power) and superposing random noise b to obtain the input of a generator;
Calculating relative position information of the aircraft, and then coding the position information by using a convergence layer to obtain a convergence matrixA i In the prediction oftAt the time of aircraft position, first willtPredicted position at time-1Converting the embedded function into a feature space to obtain featuresWill beA i Andthidden features of the aircraft at time-1Using the MLP coding result as the hidden feature of the decoding layer, and comparing the hidden feature with the feature of the decoding layerAre input into a decoding layer together to obtainFinally, obtaining the predicted flight path through MLP function coding;
Coding the result of the generator to obtain hidden features, decoding the hidden features through an MLP function, and calculating an output result by using a Softmax classifier to obtain an output score;
the loss is calculated by a loss function and the model is trained by inverse transfer gradient tuning parameters.
5. The method for data-based washing and generating fuzzy evidence track association of countermeasure network as claimed in claim 1, wherein said performing track association of track data before interruption and track data predicted bidirectionally by using fuzzy evidence interruption track association algorithm comprises:
determining a fuzzy factor set;
performing track association by using a normal type membership function, and adjusting the spread of position, speed and acceleration factors;
calculating the correlation degree of a certain flight path before interruption and a certain flight path after interruptionf ij (l) And based on a first targetnOf a track and a second targetmEach track, constituting a fuzzy correlation matrix at a time:
If the largest elementf ij (l) Greater than a set relevancy thresholdεThen flight pathiAndjthe association is successful; otherwise, the track is determinediAndjthe association fails.
6. The method for associating the track of the fuzzy evidence based on the data cleaning and the generation of the countermeasure network according to claim 1, wherein after a preliminary result is obtained by performing the track association on the track data before interruption and the track data of the bidirectional prediction by using a fuzzy evidence interruption track association algorithm, the result is corrected by combining an evidence theory.
7. The method of claim 6, wherein the correcting the result based on the combined evidence theory comprises:
setting the identification frame as aθ 1 ,θ 2 ,θ 3 },θ 1 ,θ 2 Andθ 3 respectively expressed as association, non-association and uncertainty, and three evidence bodies are constructed according to the position, the speed and the acceleration, namely:
calculating similarity between evidence bodiesc(m i ,m j );
Calculating respective evidence bodiesm i Supported degree of (c) ((c))m i );
For each evidence bodym i The supported degree is normalized to obtain evidence discountw(m i );
Evidence-based discountsw(m i ) Carrying out weighted synthesis on the evidence;
according to D-S evidence theory on evidence bodymTo carry outn1 synthesis to obtain the final combined resultm(θ i ),i(= 1, 2, 3) whenm(θ 1 )- m(θ 2 )> m(θ 3 ) The samples are considered to be associated.
8. The method for associating fuzzy evidence tracks based on data washing and generation countermeasure network as claimed in claim 1, wherein said updating the parameters of track association algorithm, making the secondary association of tracks comprises:
adding newly received flight path data, and further expanding the flight path data after processing;
membership function to second fuzzy factorμ 2 Updating to make it to the interrupt timetIs more sensitive, the membership functionμ 2 Expressed as:
in the formula (I), the compound is shown in the specification,ν (t) Representing the time influencing factor in the second blurring factor,σ x 、σ y 、σ z the spread of the position ambiguity factor is represented,τ 2 in order to adjust the degree of the adjustment,u 2 is a blurring factor representing speed;
according to interrupt time intervaltIs set with a correlation thresholdεThe relevance threshold is expressed as:ε=1- f (t);
after the track association algorithm parameters are updated, the track association algorithm is interrupted by the fuzzy evidence again to carry out track association on the track data before interruption and the two-way predicted track data.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210978480.3A CN115048371A (en) | 2022-08-16 | 2022-08-16 | Fuzzy evidence track association method based on data cleaning and generation of countermeasure network |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210978480.3A CN115048371A (en) | 2022-08-16 | 2022-08-16 | Fuzzy evidence track association method based on data cleaning and generation of countermeasure network |
Publications (1)
Publication Number | Publication Date |
---|---|
CN115048371A true CN115048371A (en) | 2022-09-13 |
Family
ID=83167212
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202210978480.3A Pending CN115048371A (en) | 2022-08-16 | 2022-08-16 | Fuzzy evidence track association method based on data cleaning and generation of countermeasure network |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN115048371A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115331488A (en) * | 2022-10-13 | 2022-11-11 | 湖南省通用航空发展有限公司 | Data processing method of general aircraft |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112465113A (en) * | 2020-11-24 | 2021-03-09 | 中国人民解放军海军航空大学 | Generating type interrupted track continuing correlation method |
WO2021202613A1 (en) * | 2020-03-31 | 2021-10-07 | Woven Planet North America, Inc. | Systems and methods for predicting agent trajectory |
CN114090718A (en) * | 2022-01-11 | 2022-02-25 | 中国人民解放军海军工程大学 | Bi-LSTM prediction and fuzzy analysis based interrupted track correlation method |
-
2022
- 2022-08-16 CN CN202210978480.3A patent/CN115048371A/en active Pending
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2021202613A1 (en) * | 2020-03-31 | 2021-10-07 | Woven Planet North America, Inc. | Systems and methods for predicting agent trajectory |
CN112465113A (en) * | 2020-11-24 | 2021-03-09 | 中国人民解放军海军航空大学 | Generating type interrupted track continuing correlation method |
CN114090718A (en) * | 2022-01-11 | 2022-02-25 | 中国人民解放军海军工程大学 | Bi-LSTM prediction and fuzzy analysis based interrupted track correlation method |
Non-Patent Citations (3)
Title |
---|
李鑫刚: "基于贝叶斯网的不确定性数据清洗", 《中国优秀硕士学位论文全文数据库-信息科技辑》 * |
王志伟等: "基于修正模糊理论和D-S证据决策的航迹关联算法", 《***仿真学报》 * |
陈玉立等: "基于注意力机制和生成对抗网络的短期航迹预测模型", 《计算机应用:"HTTPS://KNS.CNKI.NET/KCMS/DETAIL/51.1307.TP.20211201.1159.004.HTML》 * |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115331488A (en) * | 2022-10-13 | 2022-11-11 | 湖南省通用航空发展有限公司 | Data processing method of general aircraft |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US11568000B2 (en) | System and method for automatic task-oriented dialog system | |
Henderson | Machine learning for dialog state tracking: A review | |
CN110929092B (en) | Multi-event video description method based on dynamic attention mechanism | |
Fu et al. | Trembr: Exploring road networks for trajectory representation learning | |
WO2020048389A1 (en) | Method for compressing neural network model, device, and computer apparatus | |
Wang et al. | Dual dynamic memory network for end-to-end multi-turn task-oriented dialog systems | |
CN113570859B (en) | Traffic flow prediction method based on asynchronous space-time expansion graph convolution network | |
Meng et al. | Jointly learning representations of nodes and attributes for attributed networks | |
US11416688B2 (en) | Learning dialogue state tracking with limited labeled data | |
CN115048371A (en) | Fuzzy evidence track association method based on data cleaning and generation of countermeasure network | |
Zhang et al. | Adaptive online incremental learning for evolving data streams | |
CN115098789B (en) | Multidimensional interest fusion recommendation method and device based on neural network and related equipment | |
CN115186147B (en) | Dialogue content generation method and device, storage medium and terminal | |
Wu et al. | GCDST: A graph-based and copy-augmented multi-domain dialogue state tracking | |
Feng et al. | Fantastic rewards and how to tame them: A case study on reward learning for task-oriented dialogue systems | |
US7895206B2 (en) | Search query categrization into verticals | |
Chen et al. | Credit: Coarse-to-fine sequence generation for dialogue state tracking | |
CN116628612A (en) | Unsupervised anomaly detection method, device, medium and equipment | |
Chamby-Diaz et al. | Dynamic correlation-based feature selection for feature drifts in data streams | |
CN112286807B (en) | Software defect positioning system based on source code file dependency relationship | |
CN112735392B (en) | Voice processing method, device, equipment and storage medium | |
CN115048425A (en) | Data screening method and device based on reinforcement learning | |
Li et al. | Adaptive real-time prediction model for short-term traffic flow uncertainty | |
Tian et al. | Fault diagnosis strategy for few shot industrial process based on data augmentation and depth information extraction | |
Sage et al. | A residual-based approach for robust random forest regression |
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 |
Application publication date: 20220913 |
|
RJ01 | Rejection of invention patent application after publication |