CN116009592A - Unmanned aerial vehicle autonomous burst prevention method based on individual similarity pigeon swarm optimization - Google Patents
Unmanned aerial vehicle autonomous burst prevention method based on individual similarity pigeon swarm optimization Download PDFInfo
- Publication number
- CN116009592A CN116009592A CN202310087554.9A CN202310087554A CN116009592A CN 116009592 A CN116009592 A CN 116009592A CN 202310087554 A CN202310087554 A CN 202310087554A CN 116009592 A CN116009592 A CN 116009592A
- Authority
- CN
- China
- Prior art keywords
- pigeon
- unmanned aerial
- aerial vehicle
- individual
- individual similarity
- 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
Images
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T10/00—Road transport of goods or passengers
- Y02T10/10—Internal combustion engine [ICE] based vehicles
- Y02T10/40—Engine management systems
Landscapes
- Control Of Position, Course, Altitude, Or Attitude Of Moving Bodies (AREA)
Abstract
The invention discloses an unmanned aerial vehicle autonomous burst prevention method based on individual similarity pigeon swarm optimization. Step one: establishing a six-degree-of-freedom model of both the burst prevention and the prevention; step two: designing a two-loop sideslip turning overload control law; step three: establishing a three-degree-of-freedom simplified model of the burst prevention and the prevention and a maneuvering action library of the unmanned aerial vehicle; step four: generating an air situation assessment function; step five: establishing a profit matrix and designing an adaptability function; step six: pigeon group optimization method design based on individual similarity; step seven: and solving an optimal strategy by using an individual similarity pigeon group optimization method. The invention can realize automatic induction of enemy actions and real-time optimal maneuver strategy selection; the dynamic avoidance of intercepting the enemy unmanned aerial vehicle can be completed, and the battlefield viability of the unmanned aerial vehicle is effectively improved; by adopting the individual similarity pigeon cluster optimization method, the global searching capability of the system can be improved, and the accuracy is improved when the optimal strategy fitness function is solved.
Description
Technical Field
The invention relates to an unmanned aerial vehicle autonomous burst prevention method based on individual similarity pigeon swarm optimization, and belongs to the field of unmanned aerial vehicle autonomous control.
Background
The unmanned aerial vehicle is a striking weapon which can be emitted flexibly and has high hit precision, strong concealment and strong maneuvering performance. In recent years, with the rapid development of information technology, unmanned aerial vehicles are becoming an important component of an information-based hit system for a sea, land and air system. Increasingly complex battlefield environments require unmanned aerial vehicles to have greater decision making capabilities, including awareness of battlefield situations, assessment of self-state, and decision making for offensive and defensive actions. However, due to system calculation and communication delay, the unmanned aerial vehicle in the combat mode cannot completely acquire the control end instruction, which requires the unmanned aerial vehicle to have certain autonomous decision making capability. In order to improve the autonomous decision making capability, it is a necessary work to study the close range autonomous burst prevention technology of unmanned aerial vehicles.
The close-range burst prevention technique of an unmanned aerial vehicle may be defined as a technique adopted by the unmanned aerial vehicle to break through the defense of an enemy anti-conduction defense system. When the unmanned aerial vehicle enters the attack target interception area, the counterattack of the enemy anti-conduction defense system is triggered. If the unmanned aerial vehicle does not have the capability of autonomously avoiding remote fire striking, when the unmanned aerial vehicle approaches an enemy, the striking is failed by adopting modes such as signal interference or induced detonation and the like. The common defense system generally adopts an early warning radar to lock the unmanned aerial vehicle, predicts the path of the unmanned aerial vehicle, and then intercepts the unmanned aerial vehicle. However, with the development of weapon systems, the development of interception defense systems with autonomous guidance is rapid, which makes the close-range burst defense of unmanned aerial vehicles become a two-party game problem.
The game burst prevention technology has long been used, but in most research scenes, unmanned aircrafts are simply described as three-degree-of-freedom models, and cannot well reflect actual operational maneuvers. By establishing a six-degree-of-freedom nonlinear model of the unmanned aerial vehicle and designing an overload autopilot of the unmanned aerial vehicle. By adopting the overload as a control factor, an unmanned aerial vehicle maneuver library is established. And then, a two-way situation observation result is introduced, a game optimal decision system is built, and a six-degree-of-freedom model motion situation is decided by adopting a three-degree-of-freedom model, so that the solving difficulty is reduced, and the decision accuracy is improved.
In addition, the game burst prevention problem is mostly described as a two-party escape problem in a maneuvering process at the present stage, whether the attack burst prevention is successful or not is to measure the off-target rate of the unmanned aerial vehicle of the defender, and the optimal burst prevention strategy is solved by adopting strategies such as matrix game and the like. But traditional matrix gaming strategies such as maximum minimum search methods. When the conditions are the same, the solving result is unique, the intelligence is lacking, and the risk of the deduction strategy predicted by the enemy exists. By introducing the intelligent optimization method, the system intelligence can be improved better, and meanwhile, the deduction conclusion is changeable and difficult to predict.
The pigeon optimization (Pigeon Inspired Optimization, PIO) method is a classical intelligent optimization method designed by simulating the behavior of pigeon clusters during homing. By introducing solar altitude information, geomagnetic field information and landmark information, the staged behavior in pigeon homing is summarized into compass operators and landmark operators, and a global optimal solution under a high-dimensional space is found, so that a plurality of problems in the fields of parameter optimization and optimal control can be effectively solved, and the method has a good effect on solving an optimal fitness function numerical solution in close-range burst prevention. However, when facing the fitness function under complex constraint, the pigeon optimization method sometimes falls into a local optimal solution, so that policy optimization is interrupted. Therefore, on the basis of a classical pigeon swarm optimization method, an individual similarity concept improvement map and compass operators are introduced, and the improved method is applied to the close-range burst prevention problem of the unmanned aerial vehicle.
In summary, the invention provides an autonomous burst prevention method of an unmanned aerial vehicle based on individual similarity pigeon swarm optimization. The method aims to automatically sense the action situation of the enemy when the attacking unmanned aerial vehicle falls into a close-range burst-prevention scene with the defending unmanned aerial vehicle, realize real-time optimal maneuver strategy selection, finish dynamic avoidance of the enemy unmanned aerial vehicle and improve the battlefield viability.
Disclosure of Invention
1. The invention aims to:
the invention provides an unmanned aerial vehicle autonomous burst prevention method based on individual similarity pigeon swarm optimization. The method aims to automatically sense the action situation of the enemy when the attacking unmanned aerial vehicle falls into a close-range burst-prevention scene with the defending unmanned aerial vehicle, realize real-time optimal maneuver strategy selection, finish dynamic avoidance of the enemy unmanned aerial vehicle and improve the battlefield viability.
2. The technical scheme is as follows:
aiming at the short-distance burst prevention problem of unmanned aerial vehicles, the invention provides an unmanned aerial vehicle autonomous burst prevention method based on individual similarity pigeon swarm optimization, which comprises the following specific steps:
step one: establishing a six-degree-of-freedom model of both burst prevention and burst prevention
Assuming that both the burst prevention devices have the same aerodynamic shape and actuator, the six-degree-of-freedom unmanned aerial vehicle model can be described as follows:
in the method, in the process of the invention,is the mass center position change rate of the unmanned aerial vehicle; />The components of the rotational acceleration vector of the aircraft body on each axis of the body coordinate system are respectively; />The pitch angle, the yaw angle and the change rate of the inclination angle of the unmanned aerial vehicle are respectively; />Scalar values, ballistic tilt angle and ballistic deflection angle change rates for airspeed vector V, respectively.
M x ,M y ,M z The components of the moment of the mass center of all external forces (including thrust) acting on the unmanned aerial vehicle on the axes of the machine body coordinate system are respectively. X, Y, Z are components along which all external forces except the thrust are projected onto the ballistic coordinate system. I x ,I y ,I z Is the momentum moment constant of the unmanned aerial vehicle. P is the engine thrust scalar value. Alpha is the attack angle of the unmanned aerial vehicle, beta is the sideslip angle of the unmanned aerial vehicle, gamma V Is the velocity tilt angle. m is the mass of the unmanned aerial vehicle, g is the gravitational acceleration, and 9.8m/s is taken 2 。
Step two: design of two-loop sideslip turning overload control law
The six-degree-of-freedom unmanned aerial vehicle is assumed to be controllable in forces other than gravity, including aerodynamic forces and thrust forces. The introduction of the overload description can control the force magnitude, defining the overload as follows:
wherein N is an overload vector, and represents the ratio of a resultant force vector N of all external forces except gravity on the unmanned aerial vehicle relative to the weight of the unmanned aerial vehicle, and the direction is consistent with the direction of the control force vector N. N is decomposable into components (N x ,n y ,n z ) The following are respectively indicated:
wherein n is x Called tangential overrun, n y ,n z Called normal overload, n y For the vertical direction of the overload, n z Is the lateral and lateral direction overload. On the upper partSubstituting the six-degree-of-freedom model of the unmanned aerial vehicle established in the first step, the overload can be expressed as follows:
setting the speed V of the unmanned aerial vehicle to be constant, and then tangential overload n x Is unchanged. Thus, a classical two-loop sideslip Turn (STT) overload autopilot can be built with an overload, whose flight control law is expressed as follows:
in delta x ,δ y ,δ z Respectively represent the input quantity of the actuator around the x axis, the y axis and the z axis under the machine body coordinate system, and gamma c Since STT control is used, γ≡0 is used for the desired tilt angle. n is n yc ,n zc Normal overload in the y-direction and normal overload in the z-direction are desired, respectively. K (K) pp ,K dg ,K pf ,K df ,K dp And the control coefficients of the channels are respectively. Thereby, an overload closed-loop control system of the unmanned aerial vehicle can be constructed, and the control flow chart of the overload closed-loop control system is shown in fig. 1.
Step three: establishing a three-degree-of-freedom simplified model of both burst prevention and unmanned aerial vehicle basic operation action library
Based on the overload definition in the second step, the six-degree-of-freedom closed-loop control process can be expressed as a 6-state model as follows:
based on the game concept, an overall block diagram of a defense tactical plan is available as shown in fig. 2. When the initial states are consistent and the overload amounts are consistent, the six-degree-of-freedom model track can be predicted by using the three-degree-of-freedom model motion track.
On the basis of this simplified model, the velocity V is made constant, and a model of n can be established z ,n y To control the amount ofAnd (5) a basic operation action library. By changing the size of the overload, the relation between the overload control quantity and the flight attitude of the unmanned aerial vehicle, namely a basic operation action library, can be obtained, as shown in table 1:
TABLE 1
Step four: generating an air situation assessment function
Step three, a basic operation action library is obtained, and in order to select proper tactical actions from the action library, a maneuvering action selection function, namely an air situation assessment function, is required to be designed so as to perform scoring assessment on each maneuvering action of the unmanned aerial vehicle, and the optimal person is selected to execute the scoring assessment; the structure of which is shown in figure 3. Wherein V is R Airspeed vector of unmanned aerial vehicle of red square, V B Is the airspeed vector of the blue unmanned aerial vehicle. θ R Is the included angle theta between the red unmanned aerial vehicle and the target line B Is the included angle between the blue unmanned aerial vehicle and the target line.
According to the included angle of airspeed vectors of the unmanned aerial vehicle of the two parties, the air situation assessment function can be described as two constituent elements: angle situation assessment index f θ And a distance situation assessment function f R The definitions of which are respectively as follows:
f R =Le -(R-r)/K (11)
wherein L is a constant coefficient, R is the Euclidean distance between two parties, R is the interception effective distance of the unmanned aerial vehicle of the enemy, and K is a sensitivity coefficient. Let l=1, k=1000, r= (r) red +r blue )/2,r red =r blue =800 m, i.e. the mean distance between the two effects. Multiplying the two exponents, taking f=f θ ×f R The greater the value as an air situation evaluation function is, the more obvious the red square advantage is; conversely, if the value is smaller, the advantage of the blue square is more obvious.
Step five: revenue matrix creation and fitness function design
S51, establishing a benefit matrix
Assuming that the game parties can observe states at any time from each other, at t s At the simulation moment, acquiring the state quantity of the self from the six-degree-of-freedom model
Assuming that the red and blue two parties have the same basic operation action library, the basic operation action library can be generated according to the third step. Setting n zi ,n yj At t s The moment horizontal overload measuring value and the moment vertical overload measuring value. Wherein n is zi ∈[n z1 ,n z2 ,...,n zk ]K is the strategy number in the basic operation action library of the lateral-direction overload; n is n yj ∈[n y1 ,n y2 ,...,n yw ]W is the number of policies in the vertical direction overload basic operation action library. Thus, the prediction result under the numerical effect in the basic operation action library can be obtained. Taking blue square as an example, if the excessive amount n is selected zi And n yj For input quantity, blue square is at time t p Post-prediction motion stateCan be expressed as follows:
in the above formula, f 3dof The model is simplified for three degrees of freedom of the unmanned aerial vehicle. The red square can be obtained by the same methodIs provided for the predicted motion state of the vehicle. Since both the red and blue have the same basic operation action library, the policy aggregation numbers of both the red and blue are the same, and are represented by m=n=k×w. At this time, the red policy is r= { R 1 ,r 2 ,...,r m The blue strategy is b= { B } 1 ,b 2 ,...,b n The scoring matrix H may be obtained according to step four as follows:
for example, if R is selected by red m Strategy, while blue Fang Xuan b n Policy, then the corresponding score may be expressed as h mn 。
S52, obtaining fitness function of both red and blue
Adopting a mixing strategy method to design fitness functions of both red and blue, and setting the mixing strategy of the red party asBlue Fang Hunge strategy is +.>Para-blue Fang Celve Q bmx When red Fang Xuanze, line i, the fitness function can be obtained as follows:
the higher the function, the more advantageous it is for the red party. For the blue party, the blue party will select the minimum benefit corresponding to the red party maximum benefit set, corresponding to the blue Fang Zuiyou policy Q bmx * This results in a blue-square fitness function:
when the blue party selects the optimal strategy, the red party will choose the mixing strategy to maximize the benefit, thus obtaining the red party fitness function:
step six: pigeon group optimization design based on individual similarity
S61, initializing pigeon group optimization method
The pigeon colony optimization method is designed by simulating the behavior of pigeon colonies in the homing process. By introducing solar altitude information, geomagnetic field information and landmark information, the staged behavior in pigeon homing is summarized into compass operators and landmark operators, and a global optimal solution under a high-dimensional space is found, so that a plurality of problems in the fields of parameter optimization and optimal control can be effectively solved, and the method has a good effect on solving an optimal fitness function numerical solution in close-range burst prevention. However, when facing the fitness function under complex constraint, the pigeon optimization method sometimes falls into a local optimal solution, so that policy optimization is interrupted. Therefore, on the basis of a classical pigeon group optimization method, an individual similarity concept improvement map and compass operators are introduced, and the optimal strategy function in the fourth step is searched by using the classical pigeon group optimization method.
Setting the population number of pigeons as N, and expressing the position and speed of each pigeon as follows:
wherein X is i ,V i Representing the position and velocity of the ith pigeon, respectively, d representing the search space dimension.
S62, designing map and compass operator based on individual similarity
At the beginning of the cycle, one pigeon was randomly selected as globally optimal individual. At the t-th cycle, the Euclidean distance of each pigeon global optimum pigeon is calculated, and is expressed as follows:
wherein dist t (i, G) represents the position vector of the ith individual in the t-th cycle and the position vector X of the globally optimal individual G Distance between them.
The aggregation index C is set to measure the aggregation degree of pigeons. The higher the C value, the higher the pigeon density representing the global optimal position. In contrast, the pigeon density in the globally optimal position is lower. The change in the aggregation index is related to an adaptive threshold. In t cycles, if the value of the ith pigeon is less than the adaptive threshold, the aggregation index C increases as follows:
where β is a threshold coefficient. The product of the average euclidean distance of the last cycle and beta constitutes the adaptive threshold for this current cycle. By adjusting the magnitude of the beta value, the strength of the constraint can be adjusted. The adaptive threshold will be accompanied by a change in pigeon density, with increasing iteration, the distance between pigeons becoming closer and the adaptive threshold decreasing.
S63, calculating individual similarity
Reflecting individual similarity concept by aggregation degree, when C is larger than set upper limit C max When this means that the individual concentration has been too high, the individual similarity is calculated at this time, and the individual similarity concept is defined as follows:
individual similarity S t (i, j) is a piecewise function, controlled by a constant M, the structure of which can be seen in FIG. 4. In the above, dist t (i, j) represents the Euclidean distance between i, j individuals at t cycles, dist t max Representing the maximum euclidean distance between individuals within the t-cycle population. If the distance between two individuals is smaller than th 1 The individual similarity is higher. Conversely, if the distance between two individuals is greater than th 2 And the representative individuals have lower similarity.
S64, random variation strategy
Based on the definition of individual similarity, the individual similarity S between the global optimal position G and the ith individual in t times of circulation can be calculated t (i, G). The pigeons with high individual similarity are updated by adopting a random variation method, and the method is as follows:
is a mutation constant, rand is a random number between (0, 1), T 1 Is the maximum iteration number of map and compass operators, L d ,U d Which are vectors of the maximum and minimum search space ranges, respectively. random generation (L) d ,U d ) Random numbers for components between the two vector search space ranges, thereby enabling individuals with high similarity to be re-entered into the search space.
S65, iteration of map and compass operators
After the similarity processing is completed, the position and the speed of the whole group are updated by using a classical pigeon optimized map and compass operators, as follows:
f is a map coefficient, repeating the above process until the maximum number of cycles T is reached 1 。
S66, landmark operator iteration
After the map and compass operators reach the maximum iteration, performing landmark operator iteration, wherein the formula is as follows:
the maximum iteration number of the landmark operator is T 2 The population number will be halved for each iteration.And the geometric center of the residual pigeon cluster at the moment t is represented. Fitness () is an objective function that evaluates position information using the result of the calculation.
Step seven: solving optimal strategy by using individual similarity pigeon optimization method
S71, optimizing and calculating a blue square fitness function (15) by using classical pigeon to obtain a blue Fang Zuiyou strategy
Let t be s Time of dayFor a certain round starting time point, according to the return state quantity of the six-degree-of-freedom models of the red and blue, predicting the next round situation, namely the elapsed time t, by using an air situation assessment function f and a basic operation action library p The subsequent benefit matrix H. And then calculating a blue square fitness function (15) by adopting a classical pigeon cluster optimization method, namely two steps of S65 and S66. After optimized searching, blue Fang Zuiyou strategy set can be obtainedThe component values in the policy set can be used as the trend weights for selecting the corresponding policies; the greater the weight, the more likely this policy will be selected.
S72, at the same time, acquiring a blue Fang Zuiyou policy set Q bmx * And then, calculating a red square fitness function (16) by adopting an individual similarity pigeon cluster optimization method designed by a strategy six. Obtaining the optimal strategy set of red square after optimizing and searchingSimilarly, each component value in the policy set represents a trend weight for selecting a corresponding policy; the greater the weight, the more likely this policy will be selected.
S73, after determining the optimal strategy sets of the two parties, selecting the strategy corresponding to the maximum weight item in the optimal strategy sets of the two parties as the control strategy of the two parties of the next round, and outputting the control strategy to the controllers of the two parties so as to control the time t of the next round p A six degree of freedom model in the model. S71-S73 are repeated until the simulation time ends.
Overall method framework diagram reference is made to fig. 5.
Step eight: output maximum and minimum method and individual similarity pigeon optimization method comparison schematic diagram
And judging whether a maximum and minimum method or a decision method for optimizing the individual similarity pigeon group is adopted.
And if the simulation time is the maximum and minimum method, judging whether the simulation time reaches the final simulation result. If yes, outputting a fitness function chart 6, a three-dimensional track simulation chart 7, a two-dimensional horizontal track simulation chart 8 and a two-dimensional vertical simulation track chart 9; otherwise, turning to the fourth step.
If the simulation time reaches the final simulation result, the method is used for determining the optimization of the individual similarity pigeon group. If yes, outputting a fitness function diagram 10, a three-dimensional track simulation diagram 11, a two-dimensional horizontal track simulation diagram 12 and a two-dimensional vertical simulation track diagram 13; otherwise, turning to the fourth step.
The unmanned aerial vehicle autonomous burst prevention method based on the individual similarity pigeon swarm optimization method has the advantages and effects that: 1. the game method for realizing the short-range autonomous burst prevention of the unmanned aerial vehicle is provided, the short-range dynamic obstacle avoidance of the unmanned aerial vehicle is realized, and the robustness is strong. 2. The game realization method of the six-degree-of-freedom nonlinear model is provided, a three-degree-of-freedom model is adopted to predict the six-degree-of-freedom model, the simulation precision is high, and the game is fit with an actual scene. 3. The improved optimization method is used for game strategy selection, has strong global optimizing capability and can improve the strategy selection accuracy. 4. And comparing the decision effect of the individual similarity pigeon optimization decision method and the basic maximum and minimum decision method.
Drawings
FIG. 1 six-degree-of-freedom model of unmanned aerial vehicle and two-loop overload pilot control flow chart
Fig. 2 is a block diagram of a flow chart for the tactical planning of unmanned aerial vehicle in the case of sudden defending
Figure 3 game projection air defense situation diagram similarity function structure diagram
FIG. 4 is a schematic diagram of similarity function structure
FIG. 5 is a block diagram of the overall process
FIG. 6 is a schematic diagram of a maximum and minimum method output fitness function
FIG. 7 is a schematic diagram of a three-dimensional motion trail of both red and blue directions by a maximum and minimum method
FIG. 8 is a diagram showing the motion trajectories of both red and blue (horizontal direction)
FIG. 9 is a diagram showing the motion trajectories of the maximum and minimum method (vertical direction)
FIG. 10 is a schematic diagram of the output fitness function of the similarity pigeon population optimization method
FIG. 11 is a schematic diagram of three-dimensional motion trajectories of red and blue sides of similarity pigeon cluster optimization method
FIG. 12 is a schematic diagram of motion trajectories (in horizontal direction) of red and blue sides of a pigeon group optimization method
FIG. 13 is a schematic diagram of motion trajectories (in vertical direction) of red and blue sides of a pigeon group optimization method
The reference numerals and symbols in the drawings are as follows:
gamma-tilt angle
δ xc ,δ yc ,δ zc -actuator control quantity
P T Balancing thrust output
δ x ,δ y ,δ z -actual output of actuator
X-unmanned aerial vehicle state
T-actuator time constant
ω x ,ω y ,ω z -angular velocity of each shaft
n y -vertical overloading
n z -lateral-direction overload
X r -state quantity of red prescription at a certain moment
X b -state quantity of blue prescription at a certain moment
S t (i, j) -individual similarity of i, j at t cycles
M-control constant
dist t max Maximum Euclidean distance in t times of circulating population
dist t (i, j) -t cycles of the Euclidean distance between individuals i, j
t-t wheel cycle
rand (0, 1) random number
V B -blue square airspeed
V R -Red Fang Kongsu
θ R -angle between red unmanned aerial vehicle and target line
θ B -included angle between blue unmanned aerial vehicle and target line
x-horizontal position
y-position in vertical direction
z-lateral position
Detailed Description
The effectiveness of the method is illustrated by the following example of optimizing autonomous burst prevention of the unmanned aerial vehicle through a specific individual similarity pigeon group. The experimental computer was configured as an Intel Core i7-12700 processor, a 2.30Ghz main frequency, 32G memory, software version MATLAB 2020a. The autonomous burst prevention method of the unmanned aerial vehicle based on individual similarity pigeon swarm optimization comprises the following specific steps:
step one: establishing a six-degree-of-freedom model of both burst prevention and burst prevention
The red party is set as the attack party and the blue party is set as the defending party. Given the initial state of unmanned aerial vehicle of both red and blue, red party [ x ] R ,y R ,z R ]=[0,350000,0]Blue square [ x ] B ,y B ,z B ]=[25000,350000,-500]. The red square speed is constant at 600m/s, and blue Fang Hengding m/s. Red Fang Chushi ballistic dip angle 0 degree and ballistic deflection angle 0 degree. The inclination angle of the blue Fang Chushi ballistic curve is 0.07 degrees, and the deflection angle of the ballistic curve is-10 degrees. Of the two (w) x ,w y ,w z ) All 0, attack angles of 0.07 degrees, and sideslip angles of 0 degrees. The simulation compensates for 0.01s and simulates a maximum threshold of 100s.
Step two: design of two-loop sideslip turning overload control law
Setting parameters of the six-degree-of-freedom model two-loop overload autopilot in the step one as follows:
K pp =0.53,K dg =0.08,K pf =0.1059,K df =2.3012,K dp =0.76
setting the control amount range limit to n y ∈[0.8,1.2],n z ∈[-0.1,0.1]And constructing a two-loop sideslip turning closed-loop control law.
Step three: establishing a three-degree-of-freedom simplified model of both burst prevention and unmanned aerial vehicle basic operation action library
And step two, simplifying the six-degree-of-freedom model of the unmanned aerial vehicle into a three-degree-of-freedom overload model through the overload amount. Setting a red and blue two-party candidate maneuvering instruction library as follows:
n y =[0.8,0.9,1,1.1,1.2]
n z =[-0.1,0,0.1]
at this time, k= 3,w =5, and both red and blue strategies are 3×5=15 in number, i.e., m=n=15.
Step four: generating an air situation assessment function
The state quantity of both red and blue obtained in the first step and the second step is adoptedAnd->And calculating an angle situation index and a distance situation index by using the formula (10) and the formula (11). The result f obtained by calculation θ And f R Multiplying the current time situation evaluation function values can be calculated.
Step five: revenue matrix creation and fitness function design
S51, establishing a benefit matrix
According to the instructions in the basic operation action library, m×n=225 possible instruction results are shared by both red and blue. Setting the prediction time length to be 8s, calculating situation functions of the future 8s under 225 strategies by using a formula (12), and generating a benefit matrix with the size of 15 multiplied by 15.
S52, red and blue fitness function design
According to the obtained profit matrix, the adaptability functions of the red and blue can be obtained:
and (5) performing strategy optimization by using the fitness function.
Step six: pigeon group optimization method design based on individual similarity
S61, initializing a pigeon cluster optimization model
Initializing an individual similarity method, and setting the pigeon cluster size to be 10. The maximum iteration number of map and compass operators is 80 times, and the maximum iteration number of landmark operators is 10 times. The number of pigeons is 15, and the position range of each pigeon is set to be 0,1]And ensuring that the sum of all dimensions is 1, and taking a normalized value to meet the constraint. Randomly generating pigeon position and velocity information, the position being known asSpeed is +.>
S62, evaluation of aggregation degree
Setting the maximum upper limit C of the aggregation index max For 10, the Euclidean distance between all individuals and the global optimal position is then calculated by using the formula (18), and when the Euclidean distance is smaller than the adaptive threshold value, the aggregation index is increased by 1. When the aggregation index is greater than the maximum upper limit, the population aggregation is considered to be greater, and at this time, individual similarity calculation is performed. If the group aggregation degree is smaller than the maximum upper limit, the updating of the map and compass operators is directly carried out. Otherwise, calculating the similarity of the individuals.
S63, calculating individual similarity
The individual similarity calculation was performed with a high degree of aggregation, and the individual similarity constant was 3.5. And (3) calculating the individual similarity between each individual and the global optimal point by using the formula (21).
S64, random variation strategy
The mutation constant was set to 0.5, and a random mutation strategy (21) was used to determine whether each individual was required to undergo mutation. When the variation threshold is greater than the random number rand, the spatial and positional information of the individual is randomly assigned.
S65, map and compass operator
Running a map and compass operator (22), designing the numerical value of the map operator to be 0.3, so that the initial iteration stage has stronger global searching capability, the later iteration stage has stronger local searching capability, and e -Ft The pigeon group position is updated. And when the operator reaches 80 times of maximum iteration times, finishing the calculation of the map and compass operators, and entering the calculation of landmark operators.
S66, landmark operator iteration
And (3) when each iteration starts, the landmark operator performs fitness arrangement on the pigeon optimization method, namely half of individuals with low fitness, and then calculates the central position of the pigeon. And according to the central position of the pigeon group, updating the position of each pigeon in the group, and recording to obtain the global optimal point. Repeating the above process until the landmark operator reaches the maximum iteration number for 10 times, and outputting the global optimal position as a strategy selection result.
Step seven: solving optimal strategy by using individual similarity pigeon optimization method
A blue Fang Zuiyou strategy set is obtained by calculating the formula (15) by a classical pigeon cluster optimization method, and then a red-square optimal strategy set is obtained by calculating the formula (16) by an individual similarity pigeon cluster optimization method according to a blue Fang Zuiyou strategy. And inputting the strategy corresponding to the maximum weight in the two-party strategy set into the corresponding six-degree-of-freedom model controller, and performing the game of the next round. The above steps are repeated until the simulation time is over.
Step eight: output maximum and minimum method and individual similarity pigeon optimization method comparison schematic diagram
For comparison, under the same initial conditions, the simulation results of the individual similarity pigeon group optimization method of the classical maximum and minimum search method are compared respectively. The simulation time length is 400s in total, and the prediction time length is 8s each time, so that the simulation round number is 50 rounds. And if the simulation time is the maximum and minimum method, judging whether the simulation time reaches the final simulation result. If yes, outputting a fitness function chart 6, a three-dimensional track simulation chart 7, a two-dimensional horizontal track simulation chart 8 and a two-dimensional vertical simulation track chart 9; otherwise, turning to the fourth step.
If the simulation time reaches the final simulation result, the method is used for determining the optimization of the individual similarity pigeon group. If yes, outputting a fitness function diagram 10, a three-dimensional track simulation diagram 11, a two-dimensional horizontal track simulation diagram 12 and a two-dimensional vertical simulation track diagram 13; otherwise, turning to the fourth step.
According to the invention, two-sided unmanned aerial vehicles with the same performance are adopted, normal overload capacity is used as control quantity, three-degree-of-freedom unmanned aerial vehicle models are used for estimating future situations of six-degree-of-freedom models, and an air situation assessment index is designed. Generating a benefit matrix according to the situation evaluation index, constructing adaptability functions of both red and blue, and solving the adaptability functions by adopting a pigeon optimization method based on individual similarity to obtain an optimal strategy for controlling the next period. The maximum and minimum searching methods are used as comparison methods, so that the superiority of the improved method is compared.
Claims (7)
1. An unmanned aerial vehicle autonomous burst prevention method based on individual similarity pigeon swarm optimization is characterized by comprising the following steps of: the method comprises the following specific steps:
step one: establishing a six-degree-of-freedom model of both burst prevention and burst prevention
Step two: design of two-loop sideslip turning overload control law
Step three: establishing a three-degree-of-freedom simplified model of both burst prevention and unmanned aerial vehicle basic operation action library
Step four: generating an air situation assessment function, and describing the air situation assessment function as two constituent elements: angle situation assessment index f θ And a distance situation assessment function f R Multiplying the two exponents, taking f=f θ ×f R As an air situation assessment function; function valueThe bigger the red square is, the obvious advantage is; conversely, if the numerical value is smaller, the advantage of the blue square is obvious;
step five: revenue matrix creation and fitness function design
S51, establishing a profit matrix: according to the instructions in the basic operation action library, the output state of the six-degree-of-freedom model is combined, a corresponding situation function can be calculated, and a benefit matrix is generated;
s52, obtaining an adaptability function of the red and blue according to the obtained profit matrix;
step six: pigeon group optimization method design based on individual similarity
S61, initializing a pigeon optimization model;
s62, evaluating aggregation degree: when the aggregation index is larger than the maximum upper limit, the group aggregation is considered to be larger, and at the moment, individual similarity calculation is carried out; if the group aggregation degree is smaller than the maximum upper limit, the map and compass operators are updated directly;
s63, calculating individual similarity;
s64, random mutation strategy: judging whether each individual needs to be mutated or not by using a random mutation strategy;
s65, iterating the map and compass operators, so that the initial iteration has stronger global searching capability and the later iteration has stronger local searching capability;
s66, landmark operator iteration: the landmark operator performs fitness arrangement on the pigeon optimization method at the beginning of each iteration, and then calculates the central position of the pigeon; according to the central position of the pigeon group, the position of each pigeon in the group is updated, and the global optimal point is recorded and obtained; repeating the above process until the landmark operator reaches the maximum iteration number, and outputting the global optimal position as a strategy selection result;
step seven: solving optimal strategy by using individual similarity pigeon optimization method
S71, calculating a blue square fitness function by using a classical pigeon cluster optimization algorithm to obtain a blue Fang Zuiyou strategy set;
s72, calculating a red square fitness function according to the blue square optimal strategy set by using the individual similarity pigeon swarm optimization algorithm designed in the step six to obtain a red Fang Zuiyou decision;
and S73, outputting the strategies of the two next rounds until the simulation time is over, and continuously cycling the steps S71-S73.
2. The unmanned aerial vehicle autonomous burst prevention method based on individual similarity pigeon swarm optimization of claim 1, wherein the method comprises the following steps: the specific process of establishing the profit matrix in step S51 is as follows: assuming that the game parties can observe states at any time from each other, at t s At the simulation moment, acquiring the state quantity of the self from the six-degree-of-freedom modelWherein (x, y, z) is the centroid position of the unmanned aerial vehicle; v is the vector scalar value of the airspeed of the unmanned aerial vehicle, θ is the trajectory dip angle of the unmanned aerial vehicle, ψ V Is the deflection angle of the trajectory of the unmanned aerial vehicle;
assuming that the red and blue parts have the same basic operation action library, generating an action library according to the third step; setting n z ,n y Respectively represent the lateral-direction overload amount and the vertical-direction overload amount, n zi ,n yj At t s Moment horizontal direction overload n z And the vertical direction overload n y Taking values in a basic operation action library; wherein, the liquid crystal display device comprises a liquid crystal display device,
n zi ∈[n z1 ,n z2 ,...,n zk ]k is the strategy number in the basic operation action library of the lateral-direction overload;
n yj ∈[n y1 ,n y2 ,...,n yw ]w is the number of strategies in the basic operation action library of the vertical overload; thus, a prediction result under the numerical effect in the basic operation action library can be obtained; taking blue square as an example, if the excessive amount n is selected zi And n yj For input quantity, blue square is at time t p Post-prediction motion stateCan be expressed as follows:
in the above formula, f 3dof The three-degree-of-freedom simplified model of the unmanned aerial vehicle is provided; the red square can be obtained by the same methodIs a predicted motion state of the vehicle; because both the red and the blue have the same basic operation action library, the total strategies of both the red and the blue are the same, and are expressed by m=n=k×w; at this time, the red policy is r= { R 1 ,r 2 ,...,r m The blue strategy is b= { B } 1 ,b 2 ,...,b n The scoring matrix H may be obtained according to step four as follows:
3. the unmanned aerial vehicle autonomous burst prevention method based on individual similarity pigeon swarm optimization of claim 1, wherein the method comprises the following steps: the specific process of the aggregation level evaluation in step S62 is as follows: randomly selecting a pigeon as a globally optimal individual at the beginning of the cycle; at the t-th cycle, the Euclidean distance of each pigeon global optimum pigeon is calculated, and is expressed as follows:
wherein the position and velocity of the ith individual are denoted as X i =[x i1 ,x i2 ,...,x id ],V i =[v i1 ,v i2 ,...,v id ]D is the individual dimension, and N is the number of pigeons in the population; dist (dist) t (i, G) represents the position vector of the ith individual in the t-th cycle and the position vector X of the globally optimal individual G A distance therebetween;
setting an aggregation index C to measure the aggregation degree of pigeons; the higher the C value, the higher the pigeon density representing the global optimal position; in contrast, the pigeon density at the globally optimal position is lower; the change in the aggregation index is related to an adaptive threshold; in t cycles, if the value of the ith pigeon is less than the adaptive threshold, the aggregation index C increases as follows:
then C=C+1
wherein β is a threshold coefficient; the product of the average Euclidean distance and beta of the previous cycle constitutes the adaptive threshold of this cycle; by adjusting the magnitude of the beta value, the strength of the constraint can be adjusted.
4. The unmanned aerial vehicle autonomous burst prevention method based on individual similarity pigeon swarm optimization according to claim 3, wherein the method comprises the following steps: the adaptive threshold will be accompanied by a change in the average density of the pigeons, with the distance between the pigeons increasing with iteration, and the adaptive threshold will decrease.
5. The unmanned aerial vehicle autonomous burst prevention method based on individual similarity pigeon swarm optimization of claim 1, wherein the method comprises the following steps: the specific process of individual similarity evaluation in step S63 is as follows: reflecting individual similarity concept by aggregation degree, when C is larger than set upper limit C max When this means that the individual concentration has been too high, the individual similarity is calculated at this time, and the individual similarity concept is defined as follows:
individual similarity S t (i, j) is a piecewise function, controlled by a constant M; in the above, dist t (i, j) represents the Euclidean distance between i, j individuals at t cycles, dist t max Representing the maximum Euclidean distance between individuals in the t-time circulation population; if the distance between two individuals is smaller than th 1 The individual similarity is higher; conversely, if the distance between two individuals is greater than th 2 And the representative individuals have lower similarity.
6. The unmanned aerial vehicle autonomous burst prevention method based on individual similarity pigeon swarm optimization of claim 1, wherein the method comprises the following steps: the specific expression of the random mutation strategy in the step S64 is as follows: based on the definition of individual similarity, the individual similarity S between the global optimal position G and the ith individual in t times of circulation can be calculated t (i, G); the pigeons with high individual similarity are updated by adopting a random variation method, and the method is as follows:
is a mutation constant, rand is a random number between (0, 1), T 1 Is the maximum iteration number of map and compass operators, L d ,U d Vectors of maximum and minimum search space ranges, respectively; random generation (L) d ,U d ) Random numbers for components between the two vector search space ranges, thereby enabling individuals with high similarity to be re-entered into the search space.
7. An individual-based method according to claim 1The unmanned aerial vehicle autonomous burst prevention method for optimizing the similarity pigeon group is characterized by comprising the following steps of: the specific process of obtaining the optimal strategy by applying the individual similarity pigeon cluster optimization method is as follows: assume at t s The moment is the starting moment of a certain round, and the air situation assessment function f and the basic operation action library are utilized to predict the situation of the next round, namely the elapsed time t, according to the return state quantity of the six-degree-of-freedom models of the red and blue p A post benefit matrix H; then adopting a classical pigeon swarm optimization method, namely calculating a blue square fitness function according to the contents in the steps S65 and S66; after optimized searching, blue Fang Zuiyou strategy set can be obtainedThe component values in the policy set can be used as the trend weights for selecting the corresponding policies; the greater the weight, the more likely this policy will be selected; at the same time, obtain blue Fang Zuiyou policy set Q bmx * Then, calculating a red square fitness function by adopting the individual similarity pigeon cluster optimization method designed in the step six; obtaining red square optimal strategy set +.>Similarly, each component value in the policy set represents a trend weight for selecting a corresponding policy; the greater the weight, the more likely this policy will be selected; after determining the optimal policy sets of the two parties, selecting the policy corresponding to the maximum weight item in the optimal policy sets of the two parties as the control policy of the two parties of the next round, and outputting the control policy to the controllers of the two parties so as to control the time t of the next round p A six-degree-of-freedom model in the model; and step seven, repeating the step until the simulation time is over. />
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310087554.9A CN116009592A (en) | 2023-01-17 | 2023-01-17 | Unmanned aerial vehicle autonomous burst prevention method based on individual similarity pigeon swarm optimization |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310087554.9A CN116009592A (en) | 2023-01-17 | 2023-01-17 | Unmanned aerial vehicle autonomous burst prevention method based on individual similarity pigeon swarm optimization |
Publications (1)
Publication Number | Publication Date |
---|---|
CN116009592A true CN116009592A (en) | 2023-04-25 |
Family
ID=86026872
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202310087554.9A Pending CN116009592A (en) | 2023-01-17 | 2023-01-17 | Unmanned aerial vehicle autonomous burst prevention method based on individual similarity pigeon swarm optimization |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN116009592A (en) |
-
2023
- 2023-01-17 CN CN202310087554.9A patent/CN116009592A/en active Pending
Similar Documents
Publication | Publication Date | Title |
---|---|---|
WO2021174765A1 (en) | Control system based on multi-unmanned-aerial-vehicle collaborative game confrontation | |
CN113791634B (en) | Multi-agent reinforcement learning-based multi-machine air combat decision method | |
CN110083971B (en) | Self-explosion unmanned aerial vehicle cluster combat force distribution method based on combat deduction | |
CN111666631A (en) | Unmanned aerial vehicle maneuvering decision method combining hesitation fuzzy and dynamic deep reinforcement learning | |
Hu et al. | Application of deep reinforcement learning in maneuver planning of beyond-visual-range air combat | |
Yang et al. | Evasive maneuver strategy for UCAV in beyond-visual-range air combat based on hierarchical multi-objective evolutionary algorithm | |
Burgin et al. | Rule-based air combat simulation | |
CN114063644B (en) | Unmanned fighter plane air combat autonomous decision-making method based on pigeon flock reverse countermeasure learning | |
CN110928329A (en) | Multi-aircraft track planning method based on deep Q learning algorithm | |
CN113741500B (en) | Unmanned aerial vehicle air combat maneuver decision-making method for intelligent predation optimization of simulated Harris eagle | |
Li et al. | Autonomous maneuver decision-making for a UCAV in short-range aerial combat based on an MS-DDQN algorithm | |
CN114492805A (en) | Air combat maneuver decision design method based on fuzzy reasoning | |
CN113625740A (en) | Unmanned aerial vehicle air combat game method based on transfer learning pigeon swarm optimization | |
CN114330115A (en) | Neural network air combat maneuver decision method based on particle swarm search | |
Yuan et al. | Research on UCAV maneuvering decision method based on heuristic reinforcement learning | |
Chen et al. | Design and verification of UAV maneuver decision simulation system based on deep q-learning network | |
Duan et al. | Autonomous maneuver decision for unmanned aerial vehicle via improved pigeon-inspired optimization | |
Xianyong et al. | Research on maneuvering decision algorithm based on improved deep deterministic policy gradient | |
CN111773722B (en) | Method for generating maneuver strategy set for avoiding fighter plane in simulation environment | |
CN116858039A (en) | Hypersonic aircraft game guidance method, system, equipment and medium | |
CN116294811A (en) | Armed helicopter fire flight cooperative mechanism analysis method based on multi-target wolf swarm algorithm | |
CN116009592A (en) | Unmanned aerial vehicle autonomous burst prevention method based on individual similarity pigeon swarm optimization | |
CN116011315A (en) | Missile escape area fast calculation method based on K-sparse self-coding SVM | |
CN117313561B (en) | Unmanned aerial vehicle intelligent decision model training method and unmanned aerial vehicle intelligent decision method | |
CN116679742B (en) | Multi-six-degree-of-freedom aircraft collaborative combat decision-making method |
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 |