CN112464557B - Flying wing unmanned aerial vehicle redundant control surface control method based on improved hybrid multi-target PSO - Google Patents
Flying wing unmanned aerial vehicle redundant control surface control method based on improved hybrid multi-target PSO Download PDFInfo
- Publication number
- CN112464557B CN112464557B CN202011294205.7A CN202011294205A CN112464557B CN 112464557 B CN112464557 B CN 112464557B CN 202011294205 A CN202011294205 A CN 202011294205A CN 112464557 B CN112464557 B CN 112464557B
- Authority
- CN
- China
- Prior art keywords
- control
- optimal solution
- particle
- moment
- control surface
- 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.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 42
- 239000002245 particle Substances 0.000 claims abstract description 155
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 57
- RZVHIXYEVGDQDX-UHFFFAOYSA-N 9,10-anthraquinone Chemical compound C1=CC=C2C(=O)C3=CC=CC=C3C(=O)C2=C1 RZVHIXYEVGDQDX-UHFFFAOYSA-N 0.000 claims abstract description 17
- 230000006870 function Effects 0.000 claims description 45
- 238000005457 optimization Methods 0.000 claims description 30
- 230000003028 elevating effect Effects 0.000 claims description 12
- 230000035772 mutation Effects 0.000 claims description 12
- 238000006880 cross-coupling reaction Methods 0.000 claims description 8
- 238000005312 nonlinear dynamic Methods 0.000 claims description 8
- 238000005265 energy consumption Methods 0.000 claims description 7
- 238000005096 rolling process Methods 0.000 claims description 7
- 230000008859 change Effects 0.000 claims description 3
- 239000011159 matrix material Substances 0.000 claims description 3
- 238000012216 screening Methods 0.000 claims description 3
- 238000009826 distribution Methods 0.000 abstract description 31
- 238000004364 calculation method Methods 0.000 abstract description 7
- 230000008878 coupling Effects 0.000 abstract description 3
- 238000010168 coupling process Methods 0.000 abstract description 3
- 238000005859 coupling reaction Methods 0.000 abstract description 3
- 230000004044 response Effects 0.000 description 10
- 101150036841 minJ gene Proteins 0.000 description 5
- 230000008569 process Effects 0.000 description 5
- 238000013461 design Methods 0.000 description 3
- 125000003275 alpha amino acid group Chemical group 0.000 description 2
- 150000001875 compounds Chemical class 0.000 description 2
- 230000007547 defect Effects 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 238000004088 simulation Methods 0.000 description 2
- NAWXUBYGYWOOIX-SFHVURJKSA-N (2s)-2-[[4-[2-(2,4-diaminoquinazolin-6-yl)ethyl]benzoyl]amino]-4-methylidenepentanedioic acid Chemical compound C1=CC2=NC(N)=NC(N)=C2C=C1CCC1=CC=C(C(=O)N[C@@H](CC(=C)C(O)=O)C(O)=O)C=C1 NAWXUBYGYWOOIX-SFHVURJKSA-N 0.000 description 1
- 230000006978 adaptation Effects 0.000 description 1
- 230000003044 adaptive effect Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000001808 coupling effect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 230000002028 premature Effects 0.000 description 1
- 238000003860 storage Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B64—AIRCRAFT; AVIATION; COSMONAUTICS
- B64C—AEROPLANES; HELICOPTERS
- B64C13/00—Control systems or transmitting systems for actuating flying-control surfaces, lift-increasing flaps, air brakes, or spoilers
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B64—AIRCRAFT; AVIATION; COSMONAUTICS
- B64F—GROUND OR AIRCRAFT-CARRIER-DECK INSTALLATIONS SPECIALLY ADAPTED FOR USE IN CONNECTION WITH AIRCRAFT; DESIGNING, MANUFACTURING, ASSEMBLING, CLEANING, MAINTAINING OR REPAIRING AIRCRAFT, NOT OTHERWISE PROVIDED FOR; HANDLING, TRANSPORTING, TESTING OR INSPECTING AIRCRAFT COMPONENTS, NOT OTHERWISE PROVIDED FOR
- B64F5/00—Designing, manufacturing, assembling, cleaning, maintaining or repairing aircraft, not otherwise provided for; Handling, transporting, testing or inspecting aircraft components, not otherwise provided for
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2111/00—Details relating to CAD techniques
- G06F2111/04—Constraint-based CAD
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2111/00—Details relating to CAD techniques
- G06F2111/06—Multi-objective optimisation, e.g. Pareto optimisation using simulated annealing [SA], ant colony algorithms or genetic algorithms [GA]
Landscapes
- Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Aviation & Aerospace Engineering (AREA)
- Medical Informatics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Artificial Intelligence (AREA)
- Automation & Control Theory (AREA)
- Software Systems (AREA)
- Transportation (AREA)
- Computer Hardware Design (AREA)
- Geometry (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Manufacturing & Machinery (AREA)
- Control Of Position, Course, Altitude, Or Attitude Of Moving Bodies (AREA)
Abstract
The invention discloses an improved hybrid multi-target PSO-based control method for redundant control surfaces of an unmanned aerial vehicle, and belongs to the technical field of calculation, calculation or counting. Aiming at the problems that the control surface comprehensive characteristics of an advanced layout flying-wing unmanned aerial vehicle are limited, strong coupling, nonlinearity and the like and the conventional linear distribution method is difficult to control accurately, the invention provides a control redundancy distribution strategy based on a hybrid multi-target particle swarm algorithm, so that the control surface characteristics of the flying-wing unmanned aerial vehicle are effectively processed, and the control surface control redundancy in flight control is solved. The control tracking precision is ensured, the control surface smooth control is realized, and the control surface control efficiency is improved.
Description
Technical Field
The invention belongs to the technology of flight mechanics and flight simulation, in particular to a control method of a redundant control surface of an unmanned flying wing plane based on an improved hybrid multi-target PSO (PARTICLE SWARM Optimization algorithm), belonging to the technical field of calculation, calculation or counting.
Background
The flying-wing layout carrier-borne unmanned aerial vehicle has better stealth capability, better aerodynamic efficiency and higher lift-drag ratio, so that the flying-wing layout becomes the preferred aerodynamic layout mode of the high-performance unmanned aerial vehicle. In order to improve the reliability of the unmanned aerial vehicle system, a plurality of groups of control surfaces are generally arranged to form redundant control. Compared with a conventional unmanned plane, the control surface with redundant configuration can cause the problem of ambiguous control surface control function. The control distribution technology is a common method for solving the problem of controlling redundant control surfaces. However, unlike the conventional unmanned aerial vehicle, the unmanned aerial vehicle with the flying wing layout has compact arrangement of control surfaces, enhances the pneumatic interaction between the control surfaces, further generates a cross coupling effect, and needs to consider the nonlinear characteristic of the deflection of the control surfaces in the flight control process, and reduces the control distribution precision by adopting a linear control method. In addition, for the problem of flight control allocation, since each control surface is responsible for different responsibilities and control characteristics, the control surfaces cannot be considered as the same kernel under different flight conditions and flight tasks, but different optimization targets should be set according to task requirements in different flight tasks. The optimization targets in the actual flight tasks are often not single, but a plurality of targets are required to be comprehensively weighted, so that the control distribution problem of the flying-wing layout carrier-borne unmanned aerial vehicle needs to consider the nonlinear characteristic to solve the multi-target comprehensive optimization problem of strong control surface coupling and scalar control constraint.
At present, regarding the problem of operation redundancy in the flight process of the aircraft at home and abroad, conventional control allocation strategies are mostly adopted, such as: the pseudo-inverse method, the chained method, the direct allocation method and the linear programming method tend to be obviously reduced in control performance when complex constraint conditions are processed against complex control objects aiming at single optimization targets or single constraint conditions. At present, most of objects aimed by the conventional control allocation strategy are linearized control surface efficiency models, however, when nonlinear characteristics of the objects are obvious, the conventional control allocation method cannot be applied or the situation that allocation accuracy is reduced so as to influence flight control stability occurs. The control surface deflection nonlinearity and the control surface cross coupling nonlinearity characteristic exist in the control surface of the flying-wing unmanned aerial vehicle, and the nonlinearity characteristic is obvious and cannot be ignored. Therefore, the conventional control allocation strategy cannot be suitable for controlling the redundant control surface of the flying wing unmanned aerial vehicle.
The intelligent multi-objective optimization algorithm is characterized in that the intelligent algorithm is introduced into the multi-objective optimization problem, and compared with the traditional algorithm, the intelligent multi-objective optimization algorithm has the characteristics of self-organization, self-adaption, self-learning and the like, and the continuity and convexity of objective functions and constraints are not strictly required in the optimization process. And (3) taking each particle as a potential control distribution optimal solution by a particle swarm algorithm in the intelligent multi-objective optimization algorithm, and selecting a proper individual and a global optimal solution according to each objective characteristic of nonlinear control distribution so as to enable the swarm particles to converge to a desired optimal solution. In addition, the particle swarm algorithm has a faster convergence rate due to parallel calculation of all particles, and the algorithm has fewer parameters to be adjusted, is easy to realize engineering, and is suitable for solving the problem of complex nonlinear control distribution of the flying-wing layout unmanned aerial vehicle.
At present, the main idea of the traditional multi-objective particle swarm optimization algorithm is to convert the multi-objective problem into a single-objective problem, such as an objective weighting algorithm, a satisfaction compromise method and a fuzzy logic method, so that the complexity of the problem can be reduced, and the time consumption is reduced. However, such algorithms do not guarantee that all performance objectives are optimal due to the trade-off between objectives. The problem of multi-target control distribution of the flying unmanned aerial vehicle needs higher tracking precision, and if the tracking precision cannot be ensured, the stability of the system flight control is reduced, and the flight safety is affected. When the particle swarm algorithm is used for processing multi-objective optimization of a complex system, the following improved algorithms mainly exist: (1) The method effectively reduces the complexity of the algorithm by decomposing the whole target space into a series of subspaces and decomposing the optimized multi-target into a series of single targets and optimizing the single targets in each partitioned space based on a decomposed multi-target particle swarm algorithm, but the uniformity of the solution set obtained by a conventional fixed weight vector generation mode depends on the Pareto front surface shape, and a part of the solution set can be lost for complex problems, so that a good weight vector adjustment mode is not easy to plan at present to obtain a better Pareto solution set; (2) Based on a non-Pareto-dominant multi-target particle swarm algorithm, the problem of high-dimensional multi-target optimization is usually solved, when the optimization target rises to four dimensions and above, the ordering efficiency of Pareto dominant is easy to be low, new dominant rules are required to be provided to increase the selection pressure to accelerate the algorithm convergence, and the algorithm is easy to increase the complexity of a system and is not suitable for a low-dimensional system.
Therefore, the invention aims to provide an improved hybrid multi-target particle swarm optimization algorithm based on precision preference, which solves the distribution rate considering control surface deflection nonlinearity and cross coupling nonlinearity, so as to overcome the defects that a conventional control distribution strategy cannot be suitable for redundant control surface control of an unmanned flying wing aircraft and the existing particle swarm algorithm cannot meet the requirement of high tracking precision of the multi-target control distribution problem of the unmanned flying wing aircraft.
Disclosure of Invention
The invention aims to overcome the defects of the background art, and provides an unmanned aerial vehicle redundant control surface control method based on improved hybrid multi-target PSO, so as to solve the redundancy problem in the flight control of the unmanned aerial vehicle, realize the technical effect of fast tracking control instructions of each control surface of the unmanned aerial vehicle, and solve the technical problems of low distribution precision and uneven control surface response in the control distribution of the unmanned aerial vehicle by the conventional multi-target particle swarm algorithm.
The invention adopts the following technical scheme for realizing the purposes of the invention:
The control method comprises the steps of designing a nonlinear dynamic inverse flight control law by considering influences of control surface deflection nonlinearity and cross coupling nonlinearity, calculating control moment coefficients of three paths of rolling, pitching and yawing according to the nonlinear dynamic inverse flight control law, constructing a virtual control instruction, constructing a multi-objective function according to virtual control instruction tracking constraint, energy consumption indexes and smooth control indexes, taking control signals formed by control surface operation variables as particles, and solving virtual control instruction tracking signals meeting a multi-objective optimization function by adopting an improved PSO algorithm;
The improved PSO algorithm initializes an individual optimal solution set to an initial population in the constraint range of the control surface operation variable value, calculates a Pareto optimal solution set according to each particle in the population, selects the particle with the smallest objective function value in the Pareto optimal solution set, calculates the selection probability weight of the selected particle according to the expected precision of each objective function, calculates the probability that the selected particle is a global optimal solution according to the selection probability weight, randomly selects one particle from the selected particle according to the probability that the selected particle is the global optimal solution as the global optimal solution, and updates the particle and the individual optimal solution set according to the selected global optimal solution.
Further, the control method of the redundant control surface of the flying wing unmanned aerial vehicle based on the improved hybrid multi-target PSO considers the influence design of the control surface deflection nonlinearity and the cross coupling nonlinearity as follows:
Wherein, delta l1、δl2、δl3 is three lifting ailerons on the left side, delta l4 is a resistance rudder on the left side, delta r1、δr2、δr3 is three lifting ailerons on the right side, delta r4 is a resistance rudder on the right side, Torque coefficients are controlled for the roll of the three lift ailerons on the left,The moment coefficient is controlled for the roll of the left drag rudder, Torque coefficients are controlled for the roll of the three lift ailerons on the right,For the roll control moment coefficient of the right drag rudder, Δc l(δl3,δl4) is the roll control moment coefficient of the left third lift aileron with the left drag rudder, Δc l(δr3,δr4) is the roll control moment coefficient of the right third lift aileron with the right drag rudder,The moment coefficients are controlled for the pitching of the three lift ailerons on the left,The pitch control moment coefficient for the left drag rudder,The moment coefficients are controlled for the pitching of the three lift ailerons on the right,For the pitch control moment coefficient of the right drag rudder, Δc m(δl3,δl4) is the pitch control moment coefficient of the left third elevating aileron and the left drag rudder, Δc m(δr3,δr4) is the pitch control moment coefficient of the right third elevating aileron and the right drag rudder, C n (δ) is the third order term coefficient, the second order term coefficient, the first order term coefficient, the constant term of the nonlinear fitting expression ,Cn(δ)=p1δ3+p2δ2+p3δ+p4,δ=δl1,δl2,δl3,δl4,δr1,δr2,δr3,δr4,p1、p2、p3、p4 of each control surface of the yaw channel, Δc n(δl3,δl4) is the yaw control moment coefficient of the right third elevating aileron and the left drag rudder, and Δc n(δr3,δr4) is the yaw control moment coefficient of the right third elevating aileron and the right drag rudder.
Further, the control method of the redundant control surface of the flying wing unmanned aerial vehicle based on the improved hybrid multi-target PSO comprises the following steps of: J 1 is a tracking error optimization target obtained by converting virtual instruction tracking constraint into virtual instruction tracking error, v is a virtual control instruction constructed by three-channel rolling, pitching and yawing control moment coefficients calculated according to a nonlinear dynamic inverse flight control law, v d is a desired virtual control instruction, C lδd、Cmδd、Cnδd is an expected value of the rolling control moment coefficient, the pitching control moment coefficient and the yawing control moment coefficient, and 2 represents the two norms of the matrix.
Further, the control method of the redundant control surface of the flying wing unmanned aerial vehicle based on the improved hybrid multi-target PSO comprises the following steps of: minJ 2=||u||2,J2 is an energy consumption optimization target, u is a control signal formed by operation variables of each control surface of the flying unmanned aerial vehicle, Is the operation variable of the three lifting ailerons on the left side,As an operating variable of the left drag rudder,Is the operation variable of the three lifting ailerons on the right side,Operating variable of the right drag rudder.
Further, the control method of the redundant control surfaces of the flying wing unmanned aerial vehicle based on the improved hybrid multi-target PSO establishes an objective function which is minJ 3=||ut-ut-1||2,ut、ut-1 as the control surface control variable at the time t and the time t-1 according to the smooth control index, Is the operation variable of three lifting ailerons at the left side of the moment t,As the operating variable of the resistance rudder on the left side at time t,Is the operation variable of the three lifting ailerons on the right side of the moment t,As the operating variable of the right drag rudder at time t,Is the operation variable of three lifting ailerons at the left side of the time t-1,As the operating variable of the left drag rudder at time t-1,Is the operation variable of the three lifting ailerons on the right side of the time t-1,Is the operating variable of the right drag rudder at time t-1.
Further, in the control method of the redundant control surfaces of the flying wing unmanned aerial vehicle based on the improved hybrid multi-target PSO, the constraint range of the value of each control surface operation variable is as follows Is the operation variable of the kth control surface,Is the maximum value and the minimum value of the k control surface operation variable,Is the change speed of the k control surface operation variable,The maximum value and the minimum value of the variation speed of the control surface operation variable are the kth control surface operation variable.
Further, according to the control method of the redundant control surface of the flying wing unmanned aerial vehicle based on the improved hybrid multi-target PSO, the expression for calculating the selection probability weight of the selected particles according to the expected precision of each objective function is as follows: For the c-th selected particle minJ c is the minimum of the c-th objective function, ε c is the desired precision of the c-th objective function, c=1, 2,3.
Still further, according to the improved hybrid multi-target PSO-based control method for the redundant control surface of the flying wing unmanned aerial vehicle, the expression for calculating the probability of the selected particle as the global optimal solution according to the selected probability weight is as follows: p c is the probability that the c-th selected particle is the globally optimal solution.
Further, the control method of the redundant control surface of the flying wing unmanned aerial vehicle based on the improved hybrid multi-target PSO comprises the following steps of: according to the expression
Updating particle speed and position, updating an individual optimal solution set according to a Pareto dominant relationship when particles in two generations of individual optimal solution sets before and after updating meet a convergence index, performing mutation operation on the particles when the particles in the two generations of individual optimal solution sets before and after updating do not meet the convergence index, wherein the position of the ith particle after the mutation operation is x' i(t+1)=xi (t+1) +delta, sorting the updated particles according to the density, screening out particles with larger density, which prevent the algorithm from sinking into the local optimal solution set, as an updated individual optimal solution set, wherein x i(t)、xi (t+1) is the position of the ith particle at the moment t and the moment t+1, v i(t)、vi (t+1) is the speed of the ith particle at the moment t and the moment t+1, p Besti (t) is the position of the ith particle in the individual optimal solution set at the moment t, G Best (t) is the position of the global optimal solution at the moment t, c 1、c2 is a learning factor, r 1、r2 is a random number between intervals of [0,1] and is a particle group inertia weight,Omega max and omega min respectively represent the maximum and minimum inertia weight coefficients, item is the current iteration number, MAXiter is the iteration termination number, m and n are index adjustment factors, x' i (t+1) is the position of the ith particle at time t+1 after mutation operation, and delta is the position superposition amountAnd r is a random number with a value of (0, 1), and sigma is a random number with a value of more than 1.
Still further, in the control method of the redundant control surface of the flying wing unmanned aerial vehicle based on the improved hybrid multi-target PSO, the convergence index isThe expression for calculating the particle density isC is an index of convergence, N is the total number of particles in the individual optimal solution set, dis i is the minimum Euclidean distance between the ith particle in the individual optimal solution set and the nearest particle in the previous generation individual optimal solution set, D (i) is the concentration of the ith particle in the individual optimal solution set, G is the number of objective functions,AndThe maximum and minimum of the g-th objective function respectively,AndAnd c is the serial number of the particle i after all the particles are ordered according to the g objective function value.
The invention adopts the technical scheme and has the following beneficial effects:
(1) The control distribution strategy is designed to solve the problem of control surface redundancy of the flying-wing unmanned aerial vehicle, the algorithm comprehensively considers control surface deflection nonlinearity and cross coupling nonlinearity, the control distribution strategy is more in line with actual control surface operating characteristics, control surface nonlinearity is improved, and stable flight of the flying-wing unmanned aerial vehicle is realized.
(2) The invention improves the traditional multi-objective particle swarm algorithm, combines the control and distribution objective requirements, dynamically adjusts the probability that particles are selected to be global optimal solutions by calculating the difference value between the optimal value of each objective function in the current population and the expected precision, accelerates the population to converge to each objective to meet the expected precision requirements by increasing the selection probability of the optimal particles of the current low-precision function, and improves the original algorithm to solve a solution set formed by a plurality of solutions to solve a single solution so as to be more suitable for the design of the control and distribution strategy; in addition, the problem of premature convergence of the algorithm is improved through particle mutation operation, and the optimization performance is improved. The algorithm can effectively improve the convergence accuracy of solving the control distribution problem.
(3) The control distribution strategy based on the mixed multi-target particle swarm adopted by the invention is not only suitable for the flying wing unmanned aerial vehicle, but also can be popularized to the design of other control distribution strategies of aircrafts with control redundancy, can effectively treat the nonlinear characteristics of the control surfaces of the aircrafts, has higher control distribution precision and is widely applicable to occasions.
Drawings
FIG. 1 is a block diagram of an unmanned flying wing aircraft flight control distribution.
Fig. 2 is a flow chart of a control allocation algorithm for the flying wing unmanned aerial vehicle.
FIG. 3 is a flow chart of the improved hybrid multi-target particle swarm algorithm calculation.
Fig. 4 to 6 are tracking response curves of roll control moment coefficient, pitch control moment coefficient, yaw control moment coefficient using a conventional multi-target particle swarm algorithm.
Fig. 7 to 9 are tracking response curves of roll control moment coefficient, pitch control moment coefficient, yaw control moment coefficient using the modified hybrid multi-target particle swarm algorithm.
Fig. 10 to 17 are response curves of three right-side elevons, three right-side drag rudders, three left-side elevons and three left-side drag rudders controlled and distributed by using a conventional multi-target particle swarm algorithm.
Fig. 18-25 are response curves for the control of the allocation of three right-side elevons, right-side drag rudders, three left-side elevons, left-side drag rudders using the modified hybrid multi-target particle swarm algorithm.
Detailed Description
The technical scheme of the invention is described in detail below with reference to the accompanying drawings.
The invention provides an improved hybrid multi-target PSO-based control distribution method for an unmanned aerial vehicle, which solves the problem of control redundancy in flight control. Fig. 1 reflects the steps of implementing flying control allocation of the flying unmanned aerial vehicle, and specifically includes the following four steps.
Step one, the flying wing unmanned plane control system calculates and outputs three-channel control moment coefficients (a roll control moment coefficient C lδ, a pitch control moment coefficient C mδ and a yaw control moment coefficient C nδ) of roll, pitch and yaw and an engine thrust control command T according to a nonlinear dynamic inverse flight control law. Let the three-channel control moment coefficient be virtual control command v, namelyWherein,
In the formula (1), delta l1、δl2、δl3 is three lifting ailerons on the left side, delta l4 is a resistance rudder on the left side, delta r1、δr2、δr3 is three lifting ailerons on the right side, delta r4 is a resistance rudder on the right side,Torque coefficients are controlled for the roll of the three lift ailerons on the left,The moment coefficient is controlled for the roll of the left drag rudder,Torque coefficients are controlled for the roll of the three lift ailerons on the right,For the roll control moment coefficient of the right drag rudder, Δc l(δl3,δl4) is the roll control moment coefficient of the left third lift aileron with the left drag rudder, Δc l(δr3,δr4) is the roll control moment coefficient of the right third lift aileron with the right drag rudder,The moment coefficients are controlled for the pitching of the three lift ailerons on the left,The pitch control moment coefficient for the left drag rudder, The moment coefficients are controlled for the pitching of the three lift ailerons on the right,For the pitch control moment coefficient of the right drag rudder, Δc m(δl3,δl4) is the pitch control moment coefficient of the left third elevating aileron and the left drag rudder, Δc m(δr3,δr4) is the pitch control moment coefficient of the right third elevating aileron and the right drag rudder, C n (δ) is the third order term coefficient, the second order term coefficient, the first order term coefficient, the constant term of the nonlinear fitting expression ,Cn(δ)=p1δ3+p2δ2+p3δ+p4,δ=δl1,δl2,δl3,δl4,δr1,δr2,δr3,δr4,p1、p2、p3、p4 of each control surface of the yaw channel, Δc n(δl3,δl4) is the yaw control moment coefficient of the right third elevating aileron and the left drag rudder, and Δc n(δr3,δr4) is the yaw control moment coefficient of the right third elevating aileron and the right drag rudder.
Step two, calculating a control signal u composed of 8 control surface operation variables distributed and output by the flying unmanned aerial vehicle by adopting a mixed multi-target particle swarm control distribution strategy according to a control moment coefficient virtual control command v,Wherein,Is the operation variable of the three lifting ailerons on the left side,As an operating variable of the left drag rudder,Is the operation variable of the three lifting ailerons on the right side,Operating variable of the right drag rudder.
The implementation mode of the control and distribution strategy of the hybrid multi-target particle swarm unmanned aerial vehicle is shown in fig. 2, and specifically comprises the following 3 steps:
step 2.1, analyzing control surface control characteristics, and considering control surface limitation, strong coupling and nonlinearity, thereby setting constraint conditions of control signals distributed by each control surface in a mixed multi-target particle group control distribution strategy as follows:
In the formula (2), the amino acid sequence of the compound, Is the maximum value and the minimum value of the k control surface operation variable,Is the change speed of the k control surface operation variable,The maximum value and the minimum value of the variation speed of the control surface operation variable are the kth control surface operation variable.
Step 2.2, selecting a mixed multi-objective optimization function according to flying control requirements of the flying unmanned aerial vehicle:
(1) In order to enable the flying wing unmanned aerial vehicle to track the control command quickly, the virtual command tracking constraint is converted into a virtual command tracking error as a tracking error optimization target J 1,
In the formula (3), v is a control moment coefficient virtual control instruction, v d is a desired virtual control instruction, C lδd、Cmδd、Cnδd is a roll control moment coefficient, a pitch control moment coefficient and a yaw control moment coefficient expected value, and I 2 represents the two norms of the matrix;
(2) In order to minimize energy consumption during the tracking control command of the flying-wing unmanned plane, an energy consumption optimization target J 2 is set:
min J2=||u||2 (4);
(3) In order to enable each control surface to be smoothly controlled without generating mutation during the tracking control instruction of the flying-wing unmanned plane, a smooth control optimization target J 3 is set:
min J3=||ut-ut-1||2 (5),
In the formula (5), the amino acid sequence of the compound, The control surface control variable is at the moment t,
The control surface control variable is at the time t-1.
The above objective is combined to obtain a mixture optimization objective j=min { J 1,J2,J3 }.
And 2.3, processing the comprehensive characteristics of the control surfaces of the flying unmanned aerial vehicle by using the improved mixed multi-target particle swarm algorithm shown in fig. 3, and solving the optimal solution of the mixed multi-target optimization function J to obtain control signals of the control surfaces of the flying unmanned aerial vehicle, wherein the method specifically comprises the following 13 steps.
Step 2.3.1, selecting particles according to a mixed multi-objective optimization function of the flying-wing unmanned aerial vehicleAs an optimization object, a group of values is selected as an initial population of the particle swarm algorithm in the constraint condition, and an individual optimal solution set P Best is initialized as an initial population { x 1,x2,…,xN }.
And 2.3.2, bringing each particle into each objective function, calculating to obtain an adaptive value of the particle, and comparing the dominant relations of all particles to obtain all non-dominant solutions, wherein all non-dominant solutions form a Pareto optimal solution set. The dominant and non-dominant solutions are defined as follows:
Any two solutions X 1 and X 2 satisfy the following relationship:
That is, in the case that all target values of X 1 are not worse than X 2, at least one target value of X 1 is better than X 2, then X 1 is considered to dominate X 2,X1 as a non-dominated solution.
And 2.3.3, storing the Pareto optimal solution set of the initial population calculated in the step 2.3.2 into an external archive, and giving the population rule modulus of the external archive particles as the initial particle count.
Step 2.3.4 selecting particles from the external archive Arch to minimize the objective function J 1,J2,J3, respectively, to be notedThe corresponding minimum objective function value obtained is noted minJ 1,minJ2,minJ3.
Step 2.3.5, calculating particles according to the optimal value minJ 1,minJ2,minJ3 of each objective function and the expected precision epsilon 1,ε2,ε3 of each objective function obtained in step 2.3.4Selecting probability weights:
step 2.3.6, according to the particles calculated in step 2.3.5 Selecting probability weights, computing particlesThe probability of being selected as the global optimal solution G Best is calculated as follows:
Step 2.3.7 from the particles according to the probability obtained in step 2.3.6 Randomly selecting a particle as a global optimal solution G Best
In step 2.3.8, in order to ensure that the algorithm has better convergence rate in the evolutionary period in the process of solving the optimal value, and has accurate optimizing capability in the later period, the inertia weight of a given particle swarm is as follows:
In the formula (9), ω max and ω min respectively represent the maximum and minimum inertia weight coefficients, item is the current iteration number, MAXiter is the iteration termination number, m and n are index adjustment factors, the inertia weight of the algorithm can be obviously changed by adjusting the adjustment factors, and the convergence rate and development efficiency of the algorithm are improved.
The velocity and position evolution formula of particle i is:
In the formula (10), v i denotes the particle speed, x i denotes the position of the current particle, p Besti denotes the position of the i-th particle in the individual optimal solution set, G Best denotes the global optimal position, c 1、c2 denotes the learning factor, and r 1、r2 denotes the random number in the interval between [0,1 ].
In step 2.3.9, since the selected global optimal solution G Best is a probability selection manner based on the optimal value of each objective function, the algorithm has a higher tendency under the condition that the expected accuracy requirement of the virtual instruction tracking error objective function is higher, in order to prevent the algorithm from sinking into the local optimal solution too early, i.e. the optimizing result does not meet the control allocation accuracy requirement algorithm but enters the convergence state, the algorithm state needs to be judged, which is specifically as follows: calculating a convergence index C between particles of the two generations of individual optimal sets before and after updating, if the convergence index is not equal to 0, indicating that the algorithm is in an evolution state, and performing step 2.3.10; if the convergence index is equal to 0, the algorithm is in a stagnation state without meeting the precision requirement, and the variation operation is needed to be carried out on the particle swarm. The convergence index calculation formula is as follows:
In equation (11), dis i is the minimum Euclidean distance between the i-th particle in the individual optimal solution set and the nearest particle in the previous generation solution set.
The mutation operation is specifically as follows, given a mutation factor for particle i:
in formula (12), r=rand (0, 1), σ > 1.
Performing mutation operation on particles in the particle group, wherein the positions of the mutated particles are as follows:
x'i(t+1)=xi(t+1)+Δ (13),
and correcting the particles which violate the constraint after mutation, specifically correcting the particles to be boundary values.
Step 2.3.10, calculating each objective function adaptation value from the updated particles obtained in step 2.3.9. And updating the individual optimal solution set P Best according to the Pareto dominant relationship.
Step 2.3.11, comparing the dominant relationship of the current external archive Arch of the updated individual optimal solution set P Best, and if the updated particles are dominant to the particles in the external archive set, deleting the dominant particles to retain the updated particles; if the updated particles are dominated by archive set particles, the updated particles are deleted. All remain if no dominance exists.
In step 2.3.12, the number of particles obtained in step 2.3.11 may be larger than the number of particles stored in the external file Arch, so it is necessary to determine whether the number of particles obtained in step 2.3.11 is larger than the number of particles stored in the external file. If not, go to step 2.3.13; and if the number of the particles is larger than the preset threshold, sorting all the particles according to the concentration degree from small to large, storing the number of the particles according to an external storage file, retaining the particles with larger concentration degree, deleting the particles with smaller concentration degree, and obtaining the updated individual optimal solution set. Through the screening of the concentration principle, the distribution of all particles in the individual optimal solution set can be more uniform, and the algorithm can be prevented from falling into the local optimal solution. The formula for the density of particles i is as follows:
In the formula (14), G is the number of objective functions, AndThe maximum and minimum of the g-th objective function respectively,AndThe g-th objective function value of two particles adjacent to the particle i after all particles are ordered according to the g-th objective function value, and c represents the serial number of the particle i after all particles are ordered according to the g-th objective function value.
And 2.3.13, judging whether an ending condition is met, namely randomly selecting a particle from the external archive Arch to judge whether each objective function reaches the control allocation precision requirement or reaches the maximum iteration number. And when the ending condition is met, taking the optimal particles as the control variables of all control surfaces of the flying wing, and when the ending condition is not met, repeating the steps 2.3.4-2.3.13.
And thirdly, sending the engine thrust control instruction T obtained in the flight control law in the first step to the accelerator delta T and sending the control surface control signal u obtained in the second step to the pneumatic control surface of the flying unmanned aerial vehicle, so as to realize the control of the flying unmanned aerial vehicle.
And fourthly, detecting the positions X, Y, Z, the speed V, the track inclination angle gamma, the track deflection angle χ, the attack angle alpha, the sideslip angle beta, the airflow rolling angle mu and the attitude angular speeds p, q and r of the flying-wing unmanned aerial vehicle in real time, and repeating the steps one to four.
And giving reference pseudo instruction input based on the theoretical process, and obtaining a control distribution simulation contrast diagram.
Comparing the tracking response curve of the conventional multi-target particle swarm algorithm shown in fig. 4 to 6 with the tracking response curve of the improved hybrid multi-target particle swarm algorithm shown in fig. 7 to 9, the improved hybrid multi-target particle swarm with the pseudo-instruction tracking accuracy is higher than that of the conventional multi-target particle swarm algorithm.
The response curves of the conventional multi-target particle swarm algorithm control distribution shown in fig. 10 to 17 and the response curves of the improved hybrid multi-target particle swarm algorithm control distribution shown in fig. 18 to 25 can be analyzed to obtain that compared with the conventional multi-target particle swarm algorithm, the improved hybrid multi-target particle swarm has better control surface response smoothness.
Claims (8)
1. The flying wing unmanned aerial vehicle redundant control surface control method based on the improved hybrid multi-target PSO is characterized in that a nonlinear dynamic inverse flight control law is designed in consideration of influences of control surface deflection nonlinearity and cross coupling nonlinearity, virtual control instructions are constructed after control moment coefficients of three channels of rolling, pitching and yawing are calculated according to the nonlinear dynamic inverse flight control law, a multi-target function is built according to virtual control instruction tracking constraint, energy consumption indexes and smooth control indexes, control signals formed by operating variables of all control surfaces are taken as particles, and a virtual control instruction tracking signal meeting the multi-target optimization function is solved by adopting an improved PSO algorithm;
Initializing an individual optimal solution set to be an initial population in the constraint range of the control surface operation variable value, calculating a Pareto optimal solution set according to each particle in the population, selecting the particle with the smallest objective function value in the Pareto optimal solution set, calculating the selection probability weight of the selected particle according to the expected precision of each objective function, calculating the probability that the selected particle is a global optimal solution according to the selection probability weight, randomly selecting one particle from the selected particle according to the probability that the selected particle is the global optimal solution as the global optimal solution, updating the particle and the individual optimal solution set according to the selected global optimal solution, and calculating the expression of the selection probability weight of the selected particle according to the expected precision of each objective function, wherein the expression is as follows: w c is the c-th selected particle, min J c is the minimum of the c-th objective function, epsilon c is the desired precision of the c-th objective function, c=1, 2,3, and the expression for calculating the probability of the selected particle being the globally optimal solution according to the selected probability weight is: p c is the probability that the c-th selected particle is the globally optimal solution.
2. The method for controlling redundant control surfaces of an unmanned flying wing aircraft based on the improved hybrid multi-target PSO according to claim 1, wherein the nonlinear dynamic inverse flight control law designed in consideration of the influence of control surface deflection nonlinearity and cross-coupling nonlinearity is as follows:
Wherein, delta l1、δl2、δl3 is three lifting ailerons on the left side, delta l4 is a resistance rudder on the left side, delta r1、δr2、δr3 is three lifting ailerons on the right side, delta r4 is a resistance rudder on the right side, Torque coefficients are controlled for the roll of the three lift ailerons on the left,The moment coefficient is controlled for the roll of the left drag rudder,Torque coefficients are controlled for the roll of the three lift ailerons on the right,For the roll control moment coefficient of the right drag rudder, Δc l(δl3,δl4) is the roll control moment coefficient of the left third lift aileron with the left drag rudder, Δc l(δr3,δr4) is the roll control moment coefficient of the right third lift aileron with the right drag rudder,The moment coefficients are controlled for the pitching of the three lift ailerons on the left,The pitch control moment coefficient for the left drag rudder,The moment coefficients are controlled for the pitching of the three lift ailerons on the right,For the pitch control moment coefficient of the right drag rudder, Δc m(δl3,δl4) is the pitch control moment coefficient of the left third elevating aileron and the left drag rudder, Δc m(δr3,δr4) is the pitch control moment coefficient of the right third elevating aileron and the right drag rudder, C n (δ) is the third order term coefficient, the second order term coefficient, the first order term coefficient, the constant term of the nonlinear fitting expression ,Cn(δ)=p1δ3+p2δ2+p3δ+p4,δ=δl1,δl2,δl3,δl4,δr1,δr2,δr3,δr4,p1、p2、p3、p4 of each control surface of the yaw channel, Δc n(δl3,δl4) is the yaw control moment coefficient of the right third elevating aileron and the left drag rudder, and Δc n(δr3,δr4) is the yaw control moment coefficient of the right third elevating aileron and the right drag rudder.
3. The method for controlling redundant control surfaces of an flying wing unmanned aerial vehicle based on the improved hybrid multi-target PSO according to claim 1, wherein the objective function established according to the virtual control instruction tracking constraint is: J 1 is a tracking error optimization target obtained by converting virtual instruction tracking constraint into virtual instruction tracking error, v is a virtual control instruction constructed by three-channel rolling, pitching and yawing control moment coefficients calculated according to a nonlinear dynamic inverse flight control law, v d is a desired virtual control instruction, C lδd、Cmδd、Cnδd is an expected value of the rolling control moment coefficient, the pitching control moment coefficient and the yawing control moment coefficient, and 2 represents the two norms of the matrix.
4. The method for controlling redundant control surfaces of an unmanned flying wing aircraft based on improved hybrid multi-target PSO according to claim 1, wherein the objective function established according to the energy consumption index is: min J 2=||u||2,J2 is an energy consumption optimization target, u is a control signal formed by operation variables of each control surface of the flying unmanned aerial vehicle, Is the operation variable of the three lifting ailerons on the left side,As an operating variable of the left drag rudder,Is the operation variable of the three lifting ailerons on the right side,Operating variable of the right drag rudder.
5. The method for controlling redundant control surfaces of an unmanned aerial vehicle with flying wings based on the improved hybrid multi-target PSO according to claim 1, wherein the objective function established according to the smoothness control index is that the control surface manipulated variable is in J 3=||ut-ut-1||2,ut、ut-1 at the time t and at the time t-1, Is the operation variable of three lifting ailerons at the left side of the moment t,As the operating variable of the resistance rudder on the left side at time t,Is the operation variable of the three lifting ailerons on the right side of the moment t,As the operating variable of the right drag rudder at time t,Is the operation variable of three lifting ailerons at the left side of the time t-1,As the operating variable of the left drag rudder at time t-1,Is the operation variable of the three lifting ailerons on the right side of the time t-1,Is the operating variable of the right drag rudder at time t-1.
6. The method for controlling redundant control surfaces of an unmanned flying wing aircraft based on improved hybrid multi-target PSO according to claim 1, wherein the control surface operation variable value-taking constraint range is Is the operation variable of the kth control surface, Is the maximum value and the minimum value of the k control surface operation variable,Is the change speed of the k control surface operation variable,The maximum value and the minimum value of the variation speed of the control surface operation variable are the kth control surface operation variable.
7. The method for controlling the redundant control surface of the flying wing unmanned aerial vehicle based on the improved hybrid multi-target PSO according to claim 1, wherein the method for updating the particles and the individual optimal solution sets according to the selected global optimal solution is as follows: according to the expressionUpdating particle speed and position, updating an individual optimal solution set according to a Pareto dominant relationship when particles in two generations of individual optimal solution sets before and after updating meet a convergence index, performing mutation operation on the particles when the particles in the two generations of individual optimal solution sets before and after updating do not meet the convergence index, wherein the position of the ith particle after the mutation operation is x' i(t+1)=xi (t+1) +delta, sorting the updated particles according to the density, screening out particles with larger density, which prevent the algorithm from sinking into the local optimal solution set, as an updated individual optimal solution set, wherein x i(t)、xi (t+1) is the position of the ith particle at the moment t and the moment t+1, v i(t)、vi (t+1) is the speed of the ith particle at the moment t and the moment t+1, p Besti (t) is the position of the ith particle in the individual optimal solution set at the moment t, G Best (t) is the position of the global optimal solution at the moment t, c 1、c2 is a learning factor, r 1、r2 is a random number between intervals of [0,1] and is a particle group inertia weight,Omega max and omega min respectively represent the maximum and minimum inertia weight coefficients, item is the current iteration number, MAXiter is the iteration termination number, m and n are index adjustment factors, x' i (t+1) is the position of the ith particle at time t+1 after mutation operation, and delta is the position superposition amountR is a random number with a value of (0, 1), and sigma is a random number with a value of more than 1.
8. The method for controlling redundant control surfaces of an unmanned flying wing aircraft based on improved hybrid multi-target PSO as set forth in claim 7, wherein said convergence index isThe expression for calculating the particle density isC is an index of convergence, N is the total number of particles in the individual optimal solution set, dis i is the minimum Euclidean distance between the ith particle in the individual optimal solution set and the nearest particle in the previous generation individual optimal solution set, D (i) is the concentration of the ith particle in the individual optimal solution set, G is the number of objective functions,AndThe maximum and minimum of the g-th objective function respectively,AndAnd c is the serial number of the particle i after all the particles are ordered according to the g objective function value.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202011294205.7A CN112464557B (en) | 2020-11-18 | 2020-11-18 | Flying wing unmanned aerial vehicle redundant control surface control method based on improved hybrid multi-target PSO |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202011294205.7A CN112464557B (en) | 2020-11-18 | 2020-11-18 | Flying wing unmanned aerial vehicle redundant control surface control method based on improved hybrid multi-target PSO |
Publications (2)
Publication Number | Publication Date |
---|---|
CN112464557A CN112464557A (en) | 2021-03-09 |
CN112464557B true CN112464557B (en) | 2024-07-02 |
Family
ID=74836580
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202011294205.7A Active CN112464557B (en) | 2020-11-18 | 2020-11-18 | Flying wing unmanned aerial vehicle redundant control surface control method based on improved hybrid multi-target PSO |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN112464557B (en) |
Families Citing this family (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113885358B (en) * | 2021-10-27 | 2023-09-26 | 中国民航管理干部学院 | Hybrid configuration fixed wing unmanned aerial vehicle maneuver simulation control law design method |
Family Cites Families (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103645636B (en) * | 2013-11-25 | 2016-01-06 | 南京航空航天大学 | A kind of PID controller parameter optimization method |
CN110703787A (en) * | 2019-10-09 | 2020-01-17 | 南京航空航天大学 | Aircraft redundancy control method based on mixed multi-target PSO algorithm of preference matrix |
-
2020
- 2020-11-18 CN CN202011294205.7A patent/CN112464557B/en active Active
Non-Patent Citations (2)
Title |
---|
"基于目标权重导向多目标粒子群的节能减排电力***优化调度";李整等;《中国电机工程学报》;第35卷;第70页 * |
复合式旋翼飞行器多目标控制分配策略;郑峰婴等;《航空学报》;第40卷(第6期);第322727-2-322727-3、322727-6-322727-7、322727-9页 * |
Also Published As
Publication number | Publication date |
---|---|
CN112464557A (en) | 2021-03-09 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN111351488B (en) | Intelligent trajectory reconstruction reentry guidance method for aircraft | |
Han et al. | Online policy iteration ADP-based attitude-tracking control for hypersonic vehicles | |
CN107479384B (en) | The non-backstepping control method of hypersonic aircraft neural network Hybrid Learning | |
CN111813146B (en) | Reentry prediction-correction guidance method based on BP neural network prediction voyage | |
CN109062055A (en) | A kind of Near Space Flying Vehicles control system based on Back-stepping robust adaptive dynamic surface | |
CN107632518B (en) | Hypersonic aircraft neural network Hybrid Learning control method based on High-gain observer | |
CN113885320B (en) | Aircraft random robust control method based on mixed quantum pigeon swarm optimization | |
CN112068594B (en) | JAYA algorithm optimization-based course control method for small unmanned helicopter | |
CN111459026A (en) | Multi-target control distribution method for tilt rotor unmanned aerial vehicle based on competition group algorithm | |
CN112464557B (en) | Flying wing unmanned aerial vehicle redundant control surface control method based on improved hybrid multi-target PSO | |
CN116859728A (en) | Ship dynamic positioning system thrust distribution method based on improved multi-target particle swarm algorithm | |
CN114637312B (en) | Unmanned aerial vehicle energy-saving flight control method and system based on intelligent deformation decision | |
CN112327926A (en) | Self-adaptive sliding mode control method for unmanned aerial vehicle formation | |
Arik et al. | Redesign of Morphing UAV for Simultaneous Improvement of Directional Stability and Maximum Lift/Drag Ratio. | |
CN114237268A (en) | Unmanned aerial vehicle strong robust attitude control method based on deep reinforcement learning | |
CN114489098A (en) | Attitude control method of aircraft and aircraft | |
CN116755328B (en) | Tilting rotor unmanned aerial vehicle transition section flight control method based on switching fuzzy model | |
CN115657479A (en) | Improved dynamic MOPSO-based control surface fault tolerance method for flying-wing unmanned aerial vehicle | |
Abdulla et al. | Roll control system design using auto tuning LQR technique | |
CN114815878B (en) | Hypersonic aircraft collaborative guidance method based on real-time optimization and deep learning | |
CN116700013A (en) | Mixed optimization method for fastest climbing track of aircraft | |
CN115685764A (en) | Task self-adaptive anti-interference tracking control method and system for variable-span aircraft | |
CN113721620B (en) | Vehicle transverse PID control method based on particle swarm-genetic hybrid algorithm | |
CN116227338A (en) | Variable wing aircraft track optimization method based on self-adaptive sparrow searching pseudo-spectrum method | |
CN114036628B (en) | Collaborative design method for variant aircraft wing span and control strategy |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant |