CN108596342A - A kind of civil aircraft driver intention judgment method - Google Patents
A kind of civil aircraft driver intention judgment method Download PDFInfo
- Publication number
- CN108596342A CN108596342A CN201810298502.5A CN201810298502A CN108596342A CN 108596342 A CN108596342 A CN 108596342A CN 201810298502 A CN201810298502 A CN 201810298502A CN 108596342 A CN108596342 A CN 108596342A
- Authority
- CN
- China
- Prior art keywords
- trust
- partition function
- pilot
- evidence
- basic
- 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
Classifications
-
- 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
Abstract
The present invention provides a kind of civil aircraft driver intention judgment methods, it is related to pilot driver field, the present invention builds the Evidence Reasoning Model of aerial mission, the associated basic trust partition function of constructor-intention, and the intention collection of expert system database is compared according to pilot's current operation, intention matching is carried out using basic trust partition function, solves the belief function of intention, the reasoning to be intended to.The present invention is due to using the inference of intention method acted based on pilot operator, the D S evidence theory algorithms of selection are mainstream inference pattern algorithm, the model realized was both simple and reliable, scheduled target can be effectively completed again, and it is imperfect, inaccurate and uncertain to cope with the information being likely to occur in various application environments, well adapt to ability, the present invention can provide accurately and reliably presentation of information in the case where needing quick decision for pilot, and carrying out correct operation resolution for pilot provides effective ensure and support.
Description
Technical field
The present invention relates to pilot driver field, especially a kind of pilot is intended to judgment method.
Background technology
Civil aircraft main task is safe, comfortable and passenger is efficiently delivered to destination, undoubtedly, is safely
The mostly important requirement of aircarrier aircraft.In recent years, from the point of view of the statistical conditions in the aviation thing occurred every time, it is most of
The first cause of aircraft accident is human error.With current technology development level, aircarrier aircraft can't be accomplished completely
Unmanned automatic driving, therefore crew --- especially with driver --- have most close also most complicated pass with flight safety
System.Facts proved that the generation of aircraft accident is influenced by many factors, but the error ratio of flight unit, about 2/3 with
On.From all previous aircraft accident analysis and summary, driver's unsafe behaviors are mainly manifested in the following aspects:1. operation or resolution
Mistake 2. neglect or error in judgement 3. the disability of handling emergent situation improper 4. unit, resource management are improper.
In above-mentioned 4 kinds of unsafe acts, aviation accident accounts for 50% again caused by misoperation or resolution mistake
More than, although difference expert is in wrong the defining of operation or resolution, upper meeting is variant, and reduction aircrew operates or make a decision mistake
It is accidentally vital to flight safety.
Invention content
For overcome the deficiencies in the prior art, pilot is helped to carry out correctly operation and resolution, the present invention proposes a kind of
The judgment method that pilot based on evidential reasoning is intended to helps pilot to carry out correctly operation and resolution, provides a kind of fly
The method of office staff's inference of intention, to reach the cooperation safe and efficient completion of pilot by showing pilot's information most deeply concerned
The purpose of aerial mission.
The technical solution adopted by the present invention to solve the technical problems includes the following steps:
Step 1:The Evidence Reasoning Model that aerial mission is inscribed when building a certain, according to distributed sensor to pilot's
Operational motion is detected, typing EiA operational motion, and pilot operator is cleared up and compressed, wherein i=1,2 ...,
K;
The pilot operator is cleared up:[a, p (a)≤H] → [nil], wherein a are any one operational motion, [nil]
Indicate that zero, p (a) is the attribute value of operational motion a, the including but not limited to height, speed of body, course, pitch angle, roll
Angle, H are the virtual value defined according to expert system, if the attribute value p (a) of operational motion a descriptions has less than what system defined
This event is classified as [nil] by valid value H;
The pilot operator action compresses are:[a, a ..., a] → [a] will occur repeatedly number 3 in 3~10s
The secondary above operational motion merges compression;
After operational motion is cleared up, is compressed, conformation identification frame Θ:Θ={ A1,A2,…Aj,…AN, wherein
Identification framework Θ is to assume when event needs to judge, it is all for this event it can happen that universal class, and in Θ
It can happen that be exclusive events, wherein AjFor the element in identification framework Θ, indicate that the event for needing to judge corresponds to
It is potential as a result, N be identification framework in element number;
Step 2:The associated basic trust partition function of constructor-intention, and compared specially according to pilot's current operation
The intention collection of family's system database, intention matching is carried out using basic trust partition function, in this step just like giving a definition:
Define 1:Basic trust partition function m is from set 2ΘTo the mapping of [0,1], A indicates any of identification framework Θ
Subset is denoted asAnd meet
In formula (1), m (Aj) it is elements AjBasic trust partition function, indicate to need the event that judges it is corresponding it is all can
The result of energy property is to AjTrusting degree,Expression does not generate empty set any degree of belief;Indicate that the sum of basic trust partition function of all elements is equal to 1, i.e. all elements AjDegree of belief
The sum of value is 1, and each elements AjTrust angle value be more than 0 be less than 1, for an either element Aj, as m (Aj)>0, then Aj
For burnt member;
Define 2:Basic trust partition function m (Aj) determined by following formula:
m(Aj)=CFj×Wj (2)
CF in formulajFor operational motion EjWith the elements A in identification framework ΘjTrust value, WjFor corresponding weight;
Utilize the synthesis m of evidence staged synthetic method two-by-two (A), the staged of the evidence two-by-two synthetic method specific steps
It is as follows:
Take two either elements under identification framework Θ, respectively E1And E2, E1And E2Corresponding basic trust partition function
Respectively m1And m2, coke is first to correspond to A respectivelyiAnd Bj, burnt member AiBasic trust partition function be m1(Ai), burnt member BjIt is basic
It is respectively m to trust partition function2(Bj), the trust apportioning cost of two burnt members is indicated by the line segment of one [0,1], two burnt members
The line segment of corresponding [0,1] carries out arbitrary division, respectively the basic trust partition function of corresponding burnt member, i.e., by burnt member AiArbitrary point
It it is i sections, it is m that each section, which is corresponding in turn to basic trust partition function,1(A1) ... m1(Ak) ... m1(Ai), burnt member BjArbitrarily it is divided into j
Section, it is m that each section, which is corresponding in turn to basic trust partition function,2(B1) ... m2(Bk) ... m2(Bj);
Corresponding each section indicates the burnt first trust apportioning cost determined by respective basic trust distribution respectively, not table
Show entire identification frameworkΘ;
The method of salary distribution that each burnt member trusts apportioning cost is synthesized as the evidence to an event, i.e., by m1As cross
Axis, m2As the longitudinal axis, the big rectangle of a 1*1 is obtained after synthesis, includes i*j small rectangles in big matrix, big rectangle is regarded as
Total basic trust distribution, horizontal line segment table show m1The basic trust apportioning cost being assigned in corresponding burnt member, vertical line segment table show m2Distribution
Basic trust apportioning cost onto corresponding burnt member, horizontal line section and the intersection of vertical line section form a rectangle, estimate for m1(Ak)m1
(Bk), it is 0 to make to be assigned to the basic trust apportioning cost in empty set, the basic trust apportioning cost that will be assigned in empty setIt abandons, has abandoned the basic trust apportioning cost being assigned in empty set
Afterwards, total trust value can be less than 1, in each trust value m1(Ai)m2(Bj) coefficient is multiplied by distributionIt is 1 to meet total trust value;
Define 3:IfThen the composition rule of evidence is two-by-two:
The conflict spectrum between each evidence is reflected, coefficient 1/ (1-K) is known as regularization factors, mIt closes(A) it is known as m1And m2
It is orthogonal and, it is also referred to as straight and, be denoted asSuch asIt is invalid, then m1And m2It is orthogonal andIt is not present;
D-S evidence theory algorithm:If m1, m2..., mnTrust partition function for n in a certain identification framework Θ is different,
Such as m1With m2It can synthesize, thenThe result of synthesis is still a belief function;If m1With m2Result after synthesis with
m3It still is able to synthesize, thenIt is still a belief function and meets associative law:
Similarly can with and so on all synthesis obtain a belief function
It can be obtained by D-S evidence theory algorithm, for multiple trust partition function m1, m2..., mn, first by first and the
Trust partition functions for two to merge according to the composition rule of evidence two-by-two, then by the trust partition function and third after merging
A trust partition function merges, and so on, trust distribution letter until all trust partition functions are merged into one
Number;
Step 3:It solves and is intended to AnBelief function Bel (An), it is one from set 2 to define belief function BelΘTo [0,1]
Mapping, and meet
Then Bel (A) is known as the belief function of A, indicates that evidence is genuine degree of belief to A;
Th(In) be expert system database in be intended to InPredetermined threshold, if Bel (In)≥Th(In) when, that is, think
The operation of pilot instantly, which corresponds to, is intended to InAnd dynamic changes man-machine interface, Bel (In) < Th (In) then think that this operation does not correspond to
It is intended to In, do not change current interface, operation instantly continued by pilot or makes the change of assistance property, I will be intended tonWith
Assistance property is identified to be shown on interface and be referred to for pilot.
The beneficial effects of the present invention are due to using the inference of intention method acted based on pilot operator, the D- of selection
S evidence theory algorithms are mainstream inference pattern algorithm, and the model realized was not only simple and reliable, but also can effectively complete scheduled mesh
Mark, and can cope with various application environments the information that is likely to occur is imperfect, inaccurate and uncertain in the middle, it well adapts to
Ability, the present invention is based on pilot operators to act progress Rapid Inference, can be pilot in the case where needing quick decision
Accurately and reliably presentation of information is provided, carrying out correct operation resolution for pilot provides effective ensure and support.
Description of the drawings
Fig. 1 is evidence E of the present invention1The basic trust partition function of different burnt members.
Fig. 2 is evidence E of the present invention2The basic trust partition function of different burnt members.
Fig. 3 is the space representation of D-S composition rules of the present invention.
Fig. 4 is the present invention equivalent method that evidence staged synthesizes two-by-two.
Specific implementation mode
Present invention will be further explained below with reference to the attached drawings and examples.
Step 1:The Evidence Reasoning Model that aerial mission is inscribed when building a certain, according to distributed sensor to pilot's
Operational motion is detected, typing EiA operational motion, and pilot operator is cleared up and compressed, wherein i=1,2 ...,
K;
The pilot operator is cleared up:[a, p (a)≤H] → [nil], wherein a are any one operational motion, [nil]
Indicate that zero, p (a) is the attribute value of operational motion a, the including but not limited to height, speed of body, course, pitch angle, roll
Angle, H are the virtual value defined according to expert system, and the cleaning of operational motion event is that effect is widest, and it acts as by subtracting
Few action event quantity carries out matched number to reduce, if the attribute value p (a) of operational motion a descriptions is defined less than system
Virtual value H this event is classified as [nil] then this action will not generate interrelating effect;
The pilot operator action compresses are:[a, a ..., a] → [a] will occur repeatedly number 3 in 3~10s
Operational motion more than secondary merges compression, and the intention that consideration single generates the operation is possible, by compressed operation and single
It is consistent to carry out the effect that operation a is generated;
After operational motion is cleared up, is compressed, conformation identification frame Θ:Θ={ A1,A2,…Aj,…AN, wherein
Identification framework Θ is to assume when event needs to judge, it is all for this event it can happen that universal class, and in Θ
It can happen that be exclusive events, wherein AjFor the element in identification framework Θ, indicate that the event for needing to judge corresponds to
It is potential as a result, N be identification framework in element number;
Step 2:The associated basic trust partition function of constructor-intention, and compared specially according to pilot's current operation
The intention collection of family's system database, intention matching is carried out using basic trust partition function, in this step just like giving a definition:
Define 1:Basic trust partition function m is from set 2ΘTo the mapping of [0,1], A indicates any of identification framework Θ
Subset is denoted asAnd meet
In formula (1), m (Aj) it is elements AjBasic trust partition function, indicate to need the event that judges it is corresponding it is all can
The result of energy property is to AjTrusting degree, basic trust partition function reflects to AjThe degree of belief size of itself,Expression does not generate empty set any degree of belief;Indicate the basic trust of all elements
The sum of partition function is equal to 1, i.e. all elements AjThe sum of trust angle value be 1, and each elements AjTrust angle value be more than 0
Less than 1, but for an event, in the basic trust partition function of the next evidences of identification framework Θ, for one
Either element Aj, as m (Aj)>0, then AjFor burnt member;
Define 2:Basic trust partition function m (Aj) determined by following formula:
m(Aj)=CFj×Wj (2)
CF in formulajFor operational motion EjWith the elements A in identification framework ΘjTrust value, WjFor corresponding weight;
Utilize the synthesis m of evidence staged synthetic method two-by-two (A), the staged of the evidence two-by-two synthetic method specific steps
It is as follows:
As shown in Figure 1-Figure 3, Fig. 1 is evidence E of the present invention1The basic trust partition function of different burnt members, Fig. 2 are the present invention
Evidence E2The basic trust partition function of different burnt members, Fig. 3 are the space representation of D-S composition rules of the present invention.
Take two either elements under identification framework Θ, respectively E1And E2, E1And E2Corresponding basic trust partition function
Respectively m1And m2, coke is first to correspond to A respectivelyiAnd Bj, burnt member AiBasic trust partition function be m1(Ai), burnt member BjIt is basic
It is respectively m to trust partition function2(Bj), the trust apportioning cost of two burnt members is indicated by the line segment of one [0,1], two burnt members
The line segment of corresponding [0,1] carries out arbitrary division, respectively the basic trust partition function of corresponding burnt member, i.e., by burnt member AiArbitrary point
It it is i sections, it is m that each section, which is corresponding in turn to basic trust partition function,1(A1) ... m1(Ak) ... m1(Ai), burnt member BjArbitrarily it is divided into j
Section, it is m that each section, which is corresponding in turn to basic trust partition function,2(B1) ... m2(Bk) ... m2(Bj);
Corresponding each section indicates the burnt first trust apportioning cost determined by respective basic trust distribution respectively, not table
Show entire identification frameworkΘ;
The method of salary distribution that each burnt member trusts apportioning cost is synthesized as the evidence to an event, i.e., by m1As cross
Axis, m2As the longitudinal axis, the big rectangle of a 1*1 is obtained after synthesis, includes i*j small rectangles in big matrix, big rectangle is regarded as
Total basic trust distribution, horizontal line segment table show m1The basic trust apportioning cost being assigned in corresponding burnt member, vertical line segment table show m2Distribution
Basic trust apportioning cost onto corresponding burnt member, horizontal line section and the intersection of vertical line section form a rectangle, estimate for m1(Ak)m1
(Bk), it is 0 to make to be assigned to the basic trust apportioning cost in empty set, the basic trust apportioning cost that will be assigned in empty setIt abandons, has abandoned the basic trust apportioning cost being assigned in empty set
Afterwards, total trust value can be less than 1, in each trust value m1(Ai)m2(Bj) coefficient is multiplied by distributionIt is 1 to meet total trust value;
Define 3:IfThen the composition rule of evidence is two-by-two:
The conflict spectrum between each evidence is reflected, coefficient 1/ (1-K) is known as regularization factors, and m closes (A) and is known as m1With
m2It is orthogonal and, it is also referred to as straight and, be denoted asSuch asIt is invalid, then m1And m2It is orthogonal andIt is not present;
D-S evidence theory algorithm:If m1, m2..., mnTrust partition function for n in a certain identification framework Θ is different,
Such as m1With m2It can synthesize, thenThe result of synthesis is still a belief function;If m1With m2Result after synthesis with
m3It still is able to synthesize, thenIt is still a belief function and meets associative law:
Similarly can with and so on all synthesis obtain a belief function
It can be obtained by D-S evidence theory algorithm, for multiple trust partition function m1, m2..., mn, first by first and the
Trust partition functions for two to merge according to the composition rule of evidence two-by-two, then by the trust partition function and third after merging
A trust partition function merges, and so on, trust distribution letter until all trust partition functions are merged into one
Number, as shown in Figure 4;
Step 3:It solves and is intended to AnBelief function Bel (An), it is one from set 2 to define belief function BelΘTo [0,1]
Mapping, and meet
Then Bel (A) is known as the belief function of A, indicates that evidence is genuine degree of belief to A;
Th(In) be expert system database in be intended to InPredetermined threshold, predetermined threshold is by expert system database root
It is given according to actual conditions, if Bel (In)≥Th(In) when, that is, think that the operation of pilot instantly corresponds to and is intended to InAnd dynamic changes
Man-machine interface, Bel (In) < Th (In) then think the not corresponding intention I of this operationn, do not change current interface, continued by pilot
It carries out operation instantly or makes the change of assistance property, I will be intended tonIt is identified to be shown on interface for flight with helping property
Member's reference.
The present embodiment is flown for finding that airport weather is less than landing standard into close and landing to above-mentioned by aircarrier aircraft
Office staff's inference of intention method further illustrates.
Inference pattern is built, conclusion evidence collection and intention collect.Assuming that the effective action of current sensor input has A, B, C tri-
It is a, respectively represent air speed reduction;Aircraft altitude declines;Aircraft landing is overweight.It is acted by above three and compares expert system
The intention collection of system, is associated with I1, I2Two different intentions, I1Representative continues to land, I2Indicate aircraft overshoot.
1. the associated basic trust partition function of constructor-intention.
I1:
EA1(CFA1(0.2,0.6,0.2)),WA1=0.4
In formula, EA1It indicates operational motion A and is intended to I1Degree of correlation, CFA1For the credible of tri- sides operational motion A
Degree, WA1It indicates operational motion A and is intended to I1Associated weight.
mA1=(0.2 × 0.4,0.6 × 0.4,0.2 × 0.4)=(0.08,0.24,0.08)
mA1Indicate operational motion A to being intended to I1Basic trust partition function, that is, operational motion A to be intended to I1Pass through
The evidence that three sides are shown.
Similarly have:
EB1(CFB1(0.5,0.4,0.1)),WB1=0.3;
mB1=(0.5 × 0.3,0.4 × 0.3,0.1 × 0.3)=(0.15,0.12,0.03);
EC1(CFC1(0.7,0.2,0.1)),WC1=0.2;
mC1=(0.7 × 0.2,0.2 × 0.2,0.1 × 0.2)=(0.14,0.04,0.02)
For being intended to I2Have:
EA2(CFA2(0.6,0.2,0.2)),WA2=0.2
In formula, EA2It indicates operational motion A and is intended to I2Degree of correlation, CFA2For the credible of tri- sides operational motion A
Degree, WA2It indicates operational motion A and is intended to I2Associated weight.
mA2=(0.6 × 0.2,0.2 × 0.2,0.2 × 0.2)=(0.12,0.04,0.04)
mA2Indicate operational motion A to being intended to I2Basic trust partition function, that is, operational motion A to be intended to I2Pass through
The evidence that three sides are shown.
Similarly have:
EB2(CFB2(0.6,0.2,0.2)),WB2=0.2;
mB2=(0.6 × 0.2,0.2 × 0.2,0.2 × 0.2)=(0.12,0.04,0.04);
EC2(CFC1(0.3,0.4,0.3)),WC2=0.5;
mC2=(0.3 × 0.5,0.4 × 0.5,0.3 × 0.5)=(0.15,0.20,0.15)
2. the Dempster composition rules according to D-S evidence theory synthesize basic trust partition function two-by-two.
First to operating A and related intention I1It is synthesized, it should be noted that in mA1, mB1, mC1, mA2, mB2, mC2In, point
There are three values, it can be understood as three of operation not ipsilateral to three basic trust partition functions of intention, such as:mA1=
(0.2 × 0.4,0.6 × 0.4,0.2 × 0.4)=(0.08,0.24,0.08), it can be understood as mA1(a1)=0.08, mA1(a2)
=0.24, mA1(a3)=0.08 illustrates side a1, a2, a3I is intended to operation A associations respectively1Degree of support.
mA1=(0.08,0.24,0.08), mA1(D)=1- (0.08+0.24+0.08)=0.6
mA1(D) indicate that operation A associations are intended to I1Uncertain degree.
mB1=(0.15,0.12,0.03), mB1(D)=1- (0.15+0.12+0.03)=0.7
Below to mA1, mB1It is synthesized:Wherein m is mA1, mB1Intermediate state is formed by when synthesis
Evidence.
Obtain intermediate state evidence m=(0.171,0.291,0.083), then with mC1It is combined:
M=(0.171,0.291,0.083), m (D)=0.455
mC1=(0.14,0.04,0.02), mC1(D)=0.8
K '=0.171 × (0.04+0.02)+0.291 × (0.14+0.02)+0.083 × (0.14+0.04)=0.07176
Below to m, mC1It is synthesized:Wherein M is m, mC1Final evidence is formed by after synthesis.
3. calculating the belief function of the intention.
Illustrate that operational motion A, B, C are intended to I under three sides of identification framework to association1Trusting degree be 0.608.
Intention threshold value TH (I defined in expert system database1)=0.6, Bel (I1) > TH (I1), illustrate operating
A is acted, B, C are intended to I when occurring1It will occur.Aircraft should continue the offline mode to land at this time, and man-machine interface should
Display or prompt aircraft should continue the pictures such as landing maneuver.
With the above method to being intended to I2Solution can obtain Bel (I2)=0.63, if the intention defined in expert system database
Threshold value TH (I2)=0.7, Bel (I2) < TH (I2), illustrate to be intended to I when occurring in operational motion A, B, C2Do not occur;If in expert
Intention threshold value TH (I defined in system database2)=0.6, Bel (I2) > TH (I2), illustrate to occur in operational motion A, B, C
Shi Yitu I2It will occur, man-machine interface should show the operation screen that aircraft should go around at this time.
After completing collision detection, two are intended to I1, I2It is likely to occur, is intended to I at this time1Trusting degree Bel (I1) <
Bel(I2), it should be to be intended to I in the display picture of man-machine interface2Based on, and it is intended to I1Dependent picture should be shown with small picture
Or assistance property prompt display.Similarly, when multiple intentions occur and being likely to occur, it also should follow to be most likely to occur and be intended to
Display mode supplemented by remaining intention picture of key frame.
Claims (1)
1. a kind of civil aircraft driver intention judgment method, it is characterised in that include the following steps:
Step 1:The Evidence Reasoning Model that aerial mission is inscribed when building a certain, the operation according to distributed sensor to pilot
Action is detected, typing EiA operational motion, and pilot operator is cleared up and compressed, wherein i=1,2 ..., K;
The pilot operator is cleared up:[a, p (a)≤H] → [nil], wherein a are any one operational motion, and [nil] is indicated
Zero, p (a) is the attribute value of operational motion a, and the including but not limited to height, speed of body, course, pitch angle, roll angle, H is
The virtual value defined according to expert system, if the attribute value p (a) of operational motion a descriptions is less than the virtual value H that system defines,
This event is classified as [nil];
The pilot operator action compresses are:[a, a ..., a] → [a], will be occurred repeatedly in 3~10s number 3 times with
On operational motion merge compression;
After operational motion is cleared up, is compressed, conformation identification frame Θ:Θ={ A1,A2,…Aj,…AN, wherein identifying frame
Frame Θ be assume when event needs to judge, it is all for this event it can happen that universal class, and in Θ may send out
Raw situation is exclusive events, wherein AjFor the element in identification framework Θ, indicate that the event for needing to judge is corresponding all
Possibility as a result, N be identification framework in element number;
Step 2:The associated basic trust partition function of constructor-intention, and expert system is compared according to pilot's current operation
It unites the intention collection of database, intention matching is carried out using basic trust partition function, in this step just like giving a definition:
Define 1:Basic trust partition function m is from set 2ΘTo the mapping of [0,1], A indicates any subset of identification framework Θ,
It is denoted asAnd meet:
In formula (1), m (Aj) it is elements AjBasic trust partition function, indicate to need the corresponding all possibilities of event that judge
Result to AjTrusting degree,Expression does not generate empty set any degree of belief;
Indicate that the sum of basic trust partition function of all elements is equal to 1, i.e. all elements AjTrust angle value
The sum of be 1, and each elements AjTrust angle value be more than 0 be less than 1, for an either element Aj, as m (Aj)>0, then AjFor
Jiao Yuan;
Define 2:Basic trust partition function m (Aj) determined by following formula:
m(Aj)=CFj×Wj (2)
CF in formulajFor operational motion EjWith the elements A in identification framework ΘjTrust value, WjFor corresponding weight;
Using the synthesis m of evidence staged synthetic method two-by-two (A), the staged synthetic method of the evidence two-by-two specific steps are such as
Under:
Take two either elements under identification framework Θ, respectively E1And E2, E1And E2Corresponding basic trust partition function difference
For m1And m2, coke is first to correspond to A respectivelyiAnd Bj, burnt member AiBasic trust partition function be m1(Ai), burnt member BjBasic trust
Partition function is respectively m2(Bj), the trust apportioning cost of two burnt members is indicated by the line segment of one [0,1], and two burnt members are corresponding
The line segment of [0,1] carry out arbitrary division, the basic trust partition function of corresponding burnt member respectively, i.e., by burnt member AiArbitrarily it is divided into i
Section, it is m that each section, which is corresponding in turn to basic trust partition function,1(A1) ... m1(Ak) ... m1(Ai), burnt member BjArbitrarily it is divided into j sections,
It is m that each section, which is corresponding in turn to basic trust partition function,2(B1) ... m2(Bk) ... m2(Bj);
The corresponding each section trust apportioning cost for indicating the coke member determined by respective basic trust distribution respectively, is not offered as whole
A identification framework Θ;
The method of salary distribution that each burnt member trusts apportioning cost is synthesized as the evidence to an event, i.e., by m1As horizontal axis, m2
As the longitudinal axis, the big rectangle of a 1*1 is obtained after synthesis, includes i*j small rectangles in big matrix, big rectangle is regarded as to total base
This trust is distributed, and horizontal line segment table shows m1The basic trust apportioning cost being assigned in corresponding burnt member, vertical line segment table show m2It is assigned to corresponding
Basic trust apportioning cost in burnt member, horizontal line section and the intersection of vertical line section form a rectangle, estimate for m1(Ak)m1(Bk),
It is 0 to make to be assigned to the basic trust apportioning cost in empty set, the basic trust apportioning cost that will be assigned in empty setIt abandons, has abandoned the basic trust apportioning cost being assigned in empty set
Afterwards, total trust value can be less than 1, in each trust value m1(Ai)m2(Bj) coefficient is multiplied by distributionIt is 1 to meet total trust value;
Define 3:IfThen the composition rule of evidence is two-by-two:
The conflict spectrum between each evidence is reflected, coefficient 1/ (1-K) is known as regularization factors, mIt closes(A) it is known as m1And m2Just
Hand over and, it is also referred to as straight and, be denoted asSuch asIt is invalid, then m1And m2It is orthogonal and
It is not present;
D-S evidence theory algorithm:If m1, m2..., mnTrust partition function, such as m for n in a certain identification framework Θ is different1With
m2It can synthesize, thenThe result of synthesis is still a belief function;If m1With m2Result after synthesis and m3It remains to
Enough synthesis, thenIt is still a belief function and meets associative law:
Similarly can with and so on all synthesis obtain a belief function
It can be obtained by D-S evidence theory algorithm, for multiple trust partition function m1, m2..., mn, first by first and second
Trust partition function to merge according to the composition rule of evidence two-by-two, then the trust partition function after merging is believed with third
Partition function is appointed to merge, and so on, until all trust partition functions are merged into a trust partition function;
Step 3:It solves and is intended to AnBelief function Bel (An), it is one from set 2 to define belief function BelΘTo reflecting for [0,1]
It penetrates, and meets
Then Bel (A) is known as the belief function of A, indicates that evidence is genuine degree of belief to A;
Th(In) be expert system database in be intended to InPredetermined threshold, if Bel (In)≥Th(In) when, that is, think pilot
Instantly operation, which corresponds to, is intended to InAnd dynamic changes man-machine interface, Bel (In) < Th (In) then think the not corresponding intention of this operation
In, do not change current interface, operation instantly continued by pilot or makes the change of assistance property, I will be intended tonWith assistance
Property be identified to be shown on interface and referred to for pilot.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810298502.5A CN108596342A (en) | 2018-04-04 | 2018-04-04 | A kind of civil aircraft driver intention judgment method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810298502.5A CN108596342A (en) | 2018-04-04 | 2018-04-04 | A kind of civil aircraft driver intention judgment method |
Publications (1)
Publication Number | Publication Date |
---|---|
CN108596342A true CN108596342A (en) | 2018-09-28 |
Family
ID=63625487
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810298502.5A Pending CN108596342A (en) | 2018-04-04 | 2018-04-04 | A kind of civil aircraft driver intention judgment method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108596342A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109871635A (en) * | 2019-03-05 | 2019-06-11 | 中国航空综合技术研究所 | Run-time scenario modeling method towards the capture of civil aircraft top-level demands |
CN112691374A (en) * | 2021-03-24 | 2021-04-23 | 中至江西智能技术有限公司 | Evidence chain generation method and device, readable storage medium and electronic equipment |
-
2018
- 2018-04-04 CN CN201810298502.5A patent/CN108596342A/en active Pending
Non-Patent Citations (4)
Title |
---|
ZHILI TANG 等: "Research on Pilot’s Intention Reasoning Method Based on D–S Evidence Theory", 《MAN–MACHINE–ENVIRONMENT SYSTEM ENGINEERING (MMESE 2017. LECTURE NOTES IN ELECTRICAL ENGINEERING)》 * |
李长云 等编: "《智能感知技术及在电气工程中的应用》", 31 May 2017, 电子科技大学出版社 * |
胡良梅 编: "《基于信息融合的图像理解方法研究》", 31 October 2007, 合肥工业大学出版社 * |
赵小敏 等编: "《土地利用变化及其生态环境效应研究》", 31 December 2006, 地质出版社 * |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109871635A (en) * | 2019-03-05 | 2019-06-11 | 中国航空综合技术研究所 | Run-time scenario modeling method towards the capture of civil aircraft top-level demands |
CN109871635B (en) * | 2019-03-05 | 2022-06-24 | 中国航空综合技术研究所 | Operation scene modeling method for capturing top-layer requirements of civil aircraft |
CN112691374A (en) * | 2021-03-24 | 2021-04-23 | 中至江西智能技术有限公司 | Evidence chain generation method and device, readable storage medium and electronic equipment |
CN112691374B (en) * | 2021-03-24 | 2021-06-22 | 中至江西智能技术有限公司 | Evidence chain generation method and device, readable storage medium and electronic equipment |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Ginnett | Crews as groups: Their formation and their leadership | |
Mosier et al. | Automation bias: Decision making and performance in high-tech cockpits | |
US11017677B2 (en) | Decision-making aid for revising a flight plan | |
Mosier et al. | Judgment and decision making by individuals and teams: issues, models, and applications | |
CA2421220C (en) | Information system and method using analyses based on object-centric longitudinal data | |
Goh et al. | Human factors analysis of accidents involving visual flight rules flight into adverse weather | |
Shappell et al. | Human error and commercial aviation accidents: A comprehensive, fine-grained analysis using HFACS | |
Kritzinger | Aircraft system safety: Assessments for initial airworthiness certification | |
Parasuraman et al. | Automation and human performance in aviation | |
CN108596342A (en) | A kind of civil aircraft driver intention judgment method | |
Che et al. | A fault tree-based approach for aviation risk analysis considering mental workload overload | |
Ud-Din et al. | Analysis of loss of control parameters for aircraft maneuvering in general aviation | |
CN109116831B (en) | Design method of mode confusion fault logic gate of human-computer interaction dynamic fault tree | |
Perrin et al. | A systemic model of ATM safety: the Integrated Risk Picture | |
Tan et al. | Therp-cream prediction method for human failure probability for air traffic control | |
Kowalsky et al. | An analysis of pilot error-related aircraft accidents | |
Wilson | The effect of automation on the frequency of task prioritization errors on commercial aircraft flight decks: an ASRS incident report study | |
Berman et al. | Performance data errors in air carrier operations: Causes and countermeasures | |
Schutte et al. | Designing to control flight crew errors | |
Sheridan et al. | A review of human-automation interaction failures and lessons learned | |
Miller et al. | A methodology for improving mode awareness in flight guidance design | |
De Keyser et al. | Human factors in aeronautics | |
Madhavan | Cockpit task management errors: an ASRS incident report study | |
Sherry et al. | Design of a Primary Flight Display to Avoid Decelerating Below the Minimum Safe Operating Speed | |
Sprengart | An approach to goal directed information management on the flight deck |
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 | ||
WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20180928 |
|
WD01 | Invention patent application deemed withdrawn after publication |