《多操纵面先进布局飞机控制分配技术研究(飞行力学,2006).doc》由会员分享，可在线阅读，更多相关《多操纵面先进布局飞机控制分配技术研究(飞行力学,2006).doc（20页珍藏版）》请在三一文库上搜索。
1、多操纵面先进布局飞机控制分配技术研究(飞行力学,2006)Multi-control surface advanced layout aircraft control allocation technology research (flight mechanics, 2006).txt not to waste new tears for old sorrow! Now everything must be experienced, except the wife. Theres no one in 100, only two out of 50. In this paper, by pmsl
2、1 contributionPDF documents may have a poor browsing experience at the WAP end. It is recommended that you choose TXT, or download the source file to the native view.Volume 24, period 1, 2006Flying forceFL IGH T D ynamicsV oMulti-control surface advanced layout aircraft control allocation technology
3、 researchZheng yong, liu Lin(flight control department, xi an 710065, shaanxi province)Abstract: this paper introduces the proposed, mathematical description and research development of redundant control quantity in flight control system. Analyses the pseudo-inverse method, concatenated chain method
4、 and dynamic allocation based on quadratic programming three methods, through a certain type of advanced layout design and simulation of aircraft control allocation to score the advantages and disadvantages of all kinds of algorithm analysis, the conclusions are presented. Key words: multifaceted; C
5、ontrol allocation; Generalized inverse. Map classification number: V 24911; V 212 document identification code: A story number: 100220853 (2006) 0120013204The introductionThe background machine focuses on the comparative study of several control distribution algorithms.The use of advanced and effici
6、ent aerodynamic layout is the development direction of the future fighter aircraft and other aircraft, it largely increases the flight control system control compared to the three kinds of control surfaces layout for aircraft flight control multiple control surfacesControl allocation problem descrip
7、tionThe control of the aircraft is improved by controlling redundancy. The traditional system provides a more flexible, reliable, and more efficient way to implement it. Because the amount of control greatly exceeds the pilots input volume, the first question is how to design. The integrated distrib
8、ution and coordination control of multi-control is the flight control systemIn the process of plane control, torque control surface is mainly used to produce force, in the control efficiency of known and to ignore the force created by the control surface deflection, the control of multiple control s
9、urfaces of three degrees of freedom available torque algebraic description defined as follows. Lets say that the expected torque of the plane is v (t), R k, and the control side deflection is u (t), R m. This is m, , k, given v of t, u of t, so u of t is equal to v(t), the control efficiency matrix
10、for k times m rank k.In the research of the early thrust vector, the concept of the NA SA langley study is the problem of control of the control of the aerodynamic balance of the single-axis or triaxial motion. But the real direct control allocation study is the United States at 20The researchers at
11、 the heart have come up with pseudo-control control coordination and the mid-90s. The Douglas space center is in the technical research and has obtained some research results. Try the white1Considering actuator position limits and actuator rate limit, are: (1) um in u (t) um ax or less or less in u
12、(t) or less ax m m or less(2) (3)The Boeing researchers carried out a flight of wingsF 215A CT I E verifies that there is too much control over the distribution of VBecause the digital control system has a reasonable approximate time differential: u (t) = u (t) -u (t - 1) tThe laboratory researchers
13、 simulated the control allocation algorithm of the base on pseudo-inverse method on F 216 of the thrust vector. In recent years, the tailless aircraft aerodynamic effect and innovation level control technology of the conveyor control allocation problem research 3, D avid L R aney waiting on ICE (Inn
14、ovas t ive Con t ro L Effecto R) verification was carried out on boardType,T is the sampling time. In combination (1) (3), the position limits of a single sampling period can be written as:U of t is less than or equal to u of t.-(4)-The development of research has promoted the research boom in contr
15、ol distribution technology.Based on pseudo-inverse control distribution simulation verification 4. This article will be based on a certain typeDate of collection: 2005201214; Revision date: 2005212212 author introduction: is yong (19752), male, macheng hubei people, engineer, masters, is mainly enga
16、ged in the flight control theory and application research; On Sunday (19632), male, shaanxi xianyang, deputy director of the Researcher, mainly engaged in the flight control system design, control theory and application research.Type, -u (t) = m ax u (t), um in - t in, u (t) = m m in um ax, u (t) +
17、t ax. With constrained control allocation problem m description into the following standard form: (5) B (t), u (t) = v - u (t) u (t), u (t) or less or less in the expected input v (t), the need to solve the output of the distributor14U (t) (control side deflection instruction), and satisfy the const
18、raint.Flying force24 volumesType (5) solution set is the hyperplane B u (t) = v (t) and u - super box body (t) u (t), u (t) or less or less of the intersection, if both solution set are based on convex set, then the intersection is convex set. There are three possible solutions to the equation: (1)
19、an infinite number of solutions. (2) the only solution: (3) no solution: under the constraint conditions, the real control can be approached.Control allocation technology researchThe methods of controlling distribution can be summarized in two categories: non-optimized distribution and optimized dis
20、tribution. The non-optimal control allocation method is mainly a series chain distribution method, and the optimal control allocation problem exists in many kinds of concrete mathematical transformation.The transformation of control allocation problem-With binding - u (t) u (t), u (t) or less or les
21、s of the control allocation problem can be converted into the following four types of control optimization problem can be solved by: (1) direct distribution: B of the controls effectiveness matrix are given and the desired torque, m d, find the control vector u and scalar makes J = maximum, and meet
22、 the vector u make indexes such as J = d lots lots B u - m minimum. (3) the minimum control: a given matrix and vector u p, B in the inequality constraints for vector u meet index J = lots u - u p lots is minimal. (4) hybrid optimization: a given matrix and vector u p, B in the inequality constraint
23、s for vector u meet index J = W 1 lots B u - m + W 2 d lots lots u - u p lots is minimal. Thus evolved different optimal control allocation method respectively, the main to include direct geometric method, generalized inverse method, based on linear programming and distribution method based on quadr
24、atic programming.Type B u = d constraint conditions, when the 1, u = u . M (2) the minimum error: a given matrix B is found in the inequality constraintDirect geometric method, also known as constraint distribution method, the earliest established in 1993 by D u rham based on torque is set (AM) S co
25、ncept 5. Through the mapping of the control subspace to the AM S subspace by the establishment of the rudder control, the geometric method is used to find the optimal solution. The advantage is the geometrical sense of view, but 100% of the expectation motion in all AM S. The disadvantage is that wi
26、th the increase of the dimension of the torque vector, the calculation quantity and the complexity of fast speed increase, it must be combined with a better geometric optimization algorithm. Control allocation solution based on linear programming is the most commonly used numerical method of simplex
27、 method, its solving sometimes need to test it out along the border Some poles, large amount of calculation.Some scholars In foreign literature will count interior point method (In terio r Po In t m ethod s) combined with simplex solving linear programming problem, but linear programming algorithm o
28、f real-time remains to be improved. This paper analyzes the pseudo-inverse method, serial chain method, and the dynamic allocation of three algorithms based on the quadratic rule.212 pseudo-inverse (G I)The pseudo inverse is the minimum of the control quantity (energy function)The control allocation
29、 method of the optimization index. Set for J = u T u cost function, constraint conditions for B u = v, define the scalar function H (u, ) = 015 + T u T u u - v (B), according to the condition of extreme value, can obtain the right inverse solution is as follows: pseudoinverse symbols represented as:
30、 A 0 = T (T) AA - 1 A.U = W- 1In the pseudo inverse method, the plane rate and position are not consideredAnd limit, lets say that the optimal target for control allocation is: J = W and 2, 2T + lots lots 2 B u - v, W as the weights of symmetric positive definite matrices. J 2O u and partial differe
31、ntial. The solution can be derived by combining this pseudo-inverse method:(bw-1) 0, vGet weighted pseudo-inverse method, usually choose a weight matrix of the diagonal positive definite matrix, the matrix elements related to manipulate the level control efficiency, in order to control surface posit
32、ion limit amplitude square as a general corresponding diagonal elements. The control status can be increased if the power values of the control with large control efficiency are appropriately added. Permissions to avoid the high efficiency of the steering surface into saturation too earlyPeople like
33、 V iring and Bo rd ignon have come up with a redistributive pseudo-inverseMethod (R P I) pseudo-inverse method and iterative redistribution (CG I), as one of the few control input than the pseudo inverse solution of the boundary into the saturated state,The residual dynamic is expected to be allocat
34、ed between the remaining unsaturated controls. If the first redistributive control is still saturated, the remaining control is redistributed multiple times.These improved pseudo-inverse methods extend the control space to a certain extent, and generally get smaller errors when the desired target is
35、 reached. Defect is still cannot achieve the goal of all the amount of torque to the sets, when the expected movement beyond to set boundary, unable to control optimization.213 chain control allocation (DC)T minus 1 u = B (BB) vThe control input is divided into m group: u = 1. um u T, the correspond
36、ing control efficiency matrix is: B = 1. m B (B), control allocation solution are obtained by the type: 1, 2 m (6) 1 2 u u + B + B. Plus B m u = v and the control process is expressed as: 1 u = sa t u1 (P 1 v)m2, 1 u = sa t u2 P 2 (v minus B 1 u)M - 1U = sa t um P m v - B uI I = 1i.In the form, B I
37、P I = I; Sa t (?) Indicates the position of the control and the speed limit.1)Make up of the advanced layout aircraft control allocation technology research15 (12)The chain control allocation structure is shown in figure 1.M in, 1 (u - u s), 2, B u = v0 WuIf u s is equal to B u s is equal to v0, the
38、n u is the optimal solution to the upper form. If W 1 is non-singular, u = u s is the only optimal solution. The simple method of steady-state solution can be converted into a static control allocation problem by using the static allocation strategy. M in s (u s -u d)us(13)In the form, u d stands fo
39、r reference control for various methods, such as choosing pseudo-inverseFigure 1 shows the chain allocation structure schematicWell, if you take u, d = 0, W s = I, you can get a pseudo inverse. The parameter selection of the algorithm: W 1 weight matrix affects steady state control,According to the
40、control efficiency of different control surfaces. W 2 affect the dynamic process of distribution, according to the actuator or manipulate the maximum deflection rate is determined, the greater the weight change velocity distribution refers to the corresponding control. The determination of the stead
41、y-state solution with pseudo-inverse method is relatively simple Single, and the calculation speed, at the same time can be as steady state, such as, least resistance with optimizing the allocation of index achieve multi-objective control. Dynamic allocation is still the problem of constrained optim
42、ization, based on pseudo-inverse method increase the rate of quadratic optimization index, complete dynamic allocation in linear filter way, make the control allocation problem has good analytical characteristics.Concatenated chain control distribution method is simple in engineering implementation,
43、 article distribution is not affected by aircraft model approximation degree and distribution characteristics of speed and use can depend on. The disadvantage is the maximum deflection speed limit of the manipulated face.214 dynamic optimal control allocation (Q P) based on quadratic programmingV (t
44、) for virtual control (expected torque), u (t) for real control(7)Quantity, T is the sampling period. First set up the following dynamic mapping: u (t) = f (v (t), u (t), v (t),.In the optimal difan distribution, the control surface rate constraint is considered, and the penalty item that increases
45、the speed limit in the target function. The mapping between u (t) and v (t) can be represented by a first-order linear filter as the actuator is unsaturated:U (t) = F u (t) + Gv (t).2Control distribution design and simulation analysisThe design of control distribution is based on a type of advanced
46、layout aircraft (the two-vertical tail, the duck-type layout with thrust vector). Choose five control surfaces to control volume, make u = ? The x, y, z, lf,? Tf T, represent the aileron, direction rudder, flat tail, front part and a tuxedo. Set the expected three-dimensional torque vector Mc(8)The
47、distribution of control based on optimization is as follows:J is equal to m in 1U (t) 8U s (t)2-In the form of u, R m is the real control; U s, R m for the expected steady state control; V, R k for virtual control; B, R k by m is the control efficiency matrix. In non-saturable conditions the (9) can
48、 be reduced to:M (1) (u (t) -u s (t), 2 + W.U tTheorem 1: suppose (9), the description type (10) points matching problem has the following solutions: u (t) = E u s (t) + F (t - t) + Gv u (t) : E = (I - GB) 2 1 W or 2 W F = (I - GB) 2 W W - 2 of 2 G = W- 1Theorem 1 shows that the optimal solution for controlling the allocation problem is a first-order linear filter. In steady st