CN108983815A - A kind of anti-interference autonomous docking control method based on the control of terminal iterative learning - Google Patents

Abstract

The present invention relates to a kind of anti-interference docking control methods of iterative learning control, include the following steps: step 1: completing the inner ring state Design of Feedback Controller of controlled device, guarantee to realize posture and Position Tracking Control.Step 2: the embodiment for determining iteration docking operation.Step 3: iterative learning controller algorithm is realized.The present invention, come apish docking operation, to realize the trajectory predictions and tracing control of docking target, independently docks the success rate controlled under the interference such as various aerodynamic interferences to improve using the method that iterative learning controls.The advantages of this method is: only needing the location information of terminal juncture, is easy to measure and obtain；Docking control all aims at fixed point every time, highly-safe；Iterative learning controller is located at control system outer layer as add-on module, changes less, is easy to implement to original control system.

Description

An anti-interference autonomous docking control method based on terminal iterative learning control

technical field

The invention relates to an anti-interference docking control method of iterative learning control, which belongs to the field of iterative learning control.

Background technique

Precise docking control plays an important role in many industries, such as aerial refueling and ship docking, etc. The success rate of docking control is often affected by various disturbances. Taking aerial refueling as an example, it will be subject to random disturbance (atmospheric turbulence) and repulsion interference (head wave effect of refueling aircraft) during the docking process. Among them, the repulsion interference means that when the controlled object (conical tube of oil receiver) is close to the docking target (taper sleeve), the docking target will be pushed and repelled (flow field, magnetic field, etc.) speed escape, resulting in docking failure. The repulsion interference makes the docking control problem change from the fixed target shooting problem to the moving target shooting problem. Usually, the inertia of the controlled object leads to slow movement speed, while the docking target is light in weight and moves fast, which leads to the maneuverability of the controlled object is usually not enough to catch up with the escaping docking target, which greatly increases the success rate of docking. reduce. This phenomenon exists in docking control tasks in various industries. When people participate in docking control, people can predict the trajectory of the docking target through their own experience, so as to achieve smooth docking. The invention adopts an iterative learning control method to imitate the docking process of a human to realize trajectory prediction and tracking control of a docking target, thereby improving the success rate of autonomous docking control under various aerodynamic disturbances. This method is very important and meaningful for future autonomous docking control.

Contents of the invention

The invention provides an anti-interference autonomous docking control method based on iterative learning control. It improves the docking success rate of autonomous docking control under the interference of repulsive force.

The docking control process faced in the present invention is as follows:

As shown in Figure 1, a typical docking control problem is taken as an example of aerial docking, in which the controlled object is the drogue on the receiving machine, and the docking target is the drogue at the end of the hose of the tanker. The tapered pipe (controlled object) is smoothly inserted into the center of the tapered sleeve (docking target) under the condition of interference.

A docking control problem can be expressed as follows: p _pr = [x _pr y _pr z _pr ] ^T represents the position vector of the controlled object, p _dr = [x _dr y _dr z _dr ] ^T is the position vector of the docking target, these two vectors They are all uniformly defined in a certain coordinate system. The position error between the controlled object and the docking target at any time t can be defined as

Wherein, Δx _dr/pr (t), Δy _dr/pr (t), and Δz _dr/pr (t) represent three components of the vector Δp _dr/pr (t) respectively. Repulsion interference is usually a function of Δp _dr/pr (t) as an independent variable, and the closer the distance (the smaller Δp _dr/pr (t)), the stronger the repulsion, which eventually leads to the failure of docking. The radial error ΔR is an important indicator for judging whether the docking is successful or not, and is usually defined as the norm of the position error components in the y-axis and z-axis directions

A successful docking requires that when the controlled object reaches the terminal time T of the docking target plane, the radial error ΔR between the two positions should be as small as possible, and its mathematical description is as follows: The moment of passing the docking target plane (Δx _dr/pr =0)

T=min _t {Δx _dr/pr (t)=0} (3)

When the radial error ΔR(T) at the terminal moment is within the specified allowable error threshold R _C , it can be considered that the docking is successful, and its mathematical definition is as follows

ΔR(T)＜R _C (4)

The above formula (4) is called the judging criterion of the docking success rate, and R _C is a constant formulated according to actual needs, and is also called the error threshold.

Under ideal conditions without external interference, the position of the docking target will eventually stabilize at an equilibrium point (denoted as ). Considering the existence of random disturbance and repulsive disturbance, the docking target will deviate from the expected equilibrium point position in actual situations, that is, the actual position p _dr (T) of the drogue sleeve at the terminal time T satisfies the following relationship

in, Indicates the position deviation caused by random disturbance, and w _bow indicates the position deviation caused by the repulsive force. The repulsion position deviation w _bow is related to the position difference Δp _dr/pr between the two, that is, the position of the docking target will deviate as the controlled object approaches. The iterative learning control strategy of the present invention is mainly to eliminate this deviation. The random position deviation w _dr is ubiquitous in nature and is mainly produced by the vibration of air, fluid or ground. Due to the randomness of w _dr , it cannot be overcome by control methods, so it should be avoided to implement docking control when w _dr is large. Take aerial docking and refueling as an example. When the weather is bad and the drogue swings randomly in the air, the docking task is almost impossible and very dangerous. Therefore, avoid aerial docking in bad weather. Therefore, in the actual docking control task, the random disturbance amplitude of the docking target is usually limited, for example, it is considered that the docking control can be performed when the following constraints are satisfied

|w _dr |≤0.5R _C (6)

In the iterative learning process, it is necessary to represent the position and error information under different iterations. Here, the right superscript (k) is used to indicate the value of a variable at the k-th docking moment. E.g: Indicates the position error during the k-th docking process, T ^(k) represents the terminal time of the k-th docking process, and ΔR ^(k) (T ^(k) ) represents the radial error during the k-th docking process. Assuming that a total of N docking attempts have been made, among which m times of docking are successful (that is, the radial error of the terminal satisfies the judgment criterion formula (4)), the docking success rate can be expressed as

In the present invention, the radial docking error ΔR ^(k) (T ^(k) ) decreases continuously with the increase of the number of iterations k through an iterative learning control method, and finally achieves docking control with a high success rate in the presence of repulsive force interference . Figure 2 shows the control block diagram of the entire system, in which A represents the controlled object, B represents the inner loop controller, and C represents iterative learning control. The iterative learning controller takes the trajectory p _pr of the controlled object and the trajectory p _dr of the docking target as input, and generates a suitable tracking position signal through iterative learning of historical data The role of the inner loop controller is to feed back the state vector x _pr of the controlled object to achieve its own stable control, and at the same time track the given reference position signal To achieve stable fixed-point tracking, the output is the control vector u _pr of the controlled object.

In order to further illustrate the technical solution of the present invention, an anti-interference autonomous docking control method based on terminal iterative learning control proposed by the present invention includes the following steps:

Step 1: Inner loop controller design ensures position tracking

Attitude and position control methods are currently very mature, and design methods such as root locus, pole configuration, and LQR can all be realized. After completing the design of the inner loop controller, under the ideal situation without disturbance, the three-dimensional vector of the target position in any given coordinate system F _T The controlled object can reach the designated target position at the terminal time T which is Due to the existence of various interferences, there will actually be a certain tracking error, so the position of the terminal at any time can be expressed as follows

in, is the tracking error vector of the controlled object caused by the disturbance. In order to meet the basic docking requirements, it is usually required that the controlled tracking error should not be too large, and it is recommended to meet at least

|w _pr |≤0.5R _C (9)

The magnitude of the tracking error vector w _pr depends on the maneuverability of the aircraft itself and the strength of external interference. In severe weather conditions, the position fluctuates greatly, or the controlled object itself has poor maneuverability, resulting in poor tracking accuracy. It is unrealistic to achieve precise docking control. Therefore, formula (9), as a constraint on the performance of the controlled object itself and external interference, is the basic condition to ensure that the docking control can be carried out successfully.

Step 2: Identify implementation options for the iterative docking process

Applying iterative learning control requires planning the entire iterative learning process, determining when to start docking, when to end docking, and where to return for the next iteration.

A typical iterative learning process can be described as follows.

(1) The controlled object first moves to an initial position at a certain distance behind the docking target (at this distance, the repulsion interference is weak enough, generally more than twice the length of the controlled object), and observes the position fluctuations of the docking target and itself Happening. If the weather and other external conditions are good, and the random position deviation of the controlled object satisfies constraints (6) and (9), it is considered that there are conditions for successful docking.

(2) Stay for a period of time (greater than 10s), use this period of time to take the average value of the trajectory data of the docking target, and obtain the equilibrium position of the controlled object

(3) The controlled object slowly approaches the target position under the action of the inner loop controller Until reaching the end time T ^(k) , then immediately return to the initial position.

(4) Record the position of the controlled object at the terminal moment of this docking docking target position Calculate the next target position by iterative learning algorithm Then repeat step (1) for the next docking attempt.

Step 3: Iterative Learning Controller Algorithm Implementation

Iterative Learning Algorithm and Controlled Object Position and docking target position as input, and outputs the desired target docking position The specific expression is as follows

in It is used to estimate the position deviation of the docking target due to repulsion interference, and its update law is as follows

and It is used to offset the tracking deviation caused by the insufficient maneuverability of the controlled object, and its update law is as follows

Two constant vectors involved in the two formulas are defined as follows

in, i=1,2,3. parameter The closer k _li is to 1, the higher the utilization rate of historical data, but the slower the iteration speed.

The advantages and beneficial effects of the present invention are: only the location information of the terminal is needed, which is easy to measure and obtain; each docking control is aimed at a fixed point, which has high safety; the iterative learning controller is an additional module located on the outer layer of the control system, The original control system has few changes and is easy to realize.

Description of drawings

Figure 1 is a schematic diagram of the docking system taking air docking as an example.

Figure 2 is a structural diagram of the docking control system.

Figure 3 is a diagram showing the iterative learning process.

Fig. 4 is a diagram of the implementation steps of the present invention.

The symbols in the figure are explained as follows:

Figure 1: The number 1 represents the cone pipe of the controlled object, 2 represents the oil receiving machine, 3 represents the docking target drogue, 4 represents the refueling machine, p _pr represents the trajectory of the controlled object, p _dr represents the trajectory of the docking target, and F _T represents Tanker coordinate system o _t x _t y _t z _t , where the directions of axes x _t , y _t , and z _t are forward, right, and vertically downward, respectively.

Figure 2: A represents the controlled object, B represents the inner loop controller, C represents iterative learning control, p _pr represents the trajectory of the controlled object, p _dr represents the trajectory of the docking target, It is the position tracking signal input to the inner loop control system, x _pr represents the state vector of the controlled object, and u _pr represents the direct control quantity of the controlled object.

Figure 3: The dotted line indicates the position of the drogue sleeve, and the solid line indicates the position of the tapered tube. The distribution of the picture from top to bottom shows the trajectory components of the cone tube and the drogue sleeve in the x, y, and z directions, and the distribution from left to right shows the trajectory data of the first to fourth docking attempts. The vertical coordinate unit is m, and the horizontal coordinate unit is s.

Detailed ways

Please refer to FIGS. 1-4 , the present invention provides an anti-jamming autonomous docking control based on terminal iterative learning control. Here, the aerial refueling simulation system is taken as an example for implementation and demonstration. The method specifically includes the following steps:

Step 1: Design the inner loop state feedback controller of the controlled object.

Attitude and position control methods are currently very mature, such as root locus, pole configuration, and LQR design methods can all be realized. Here we use the LQR method to design the inner loop controller of the aircraft itself. After completing the design of the inner loop controller, the aircraft can stably track the three-dimensional vector of the given target position

Step 2: Determine the implementation of the iterative docking process.

The docking process is stipulated as follows:

(1) With the initial position 6 meters directly behind the drogue sleeve, the conical pipe of the oil receiver is behind the drogue sleeve to observe the fluctuation of the position of the drogue sleeve and itself, and the random disturbance is set to |w _dr |≤0.2R _C , |w _pr |≤0.2 R _C satisfies the docking conditions of formula (6) and formula (9).

(2) Wait for 20s, record the track data of the taper sleeve and get the balance position of the taper sleeve

(3) The tapered tube approaches the predicted target position of the tapered sleeve along the x direction at a constant speed of 0.5m/s Until the terminal moment, then take R _C =0.15m to judge whether the docking is successful and return to the initial position immediately.

(4) Record the position of the cone tube and the cone sleeve at the terminal moment, and calculate the next aiming position

Taking the first docking attempt (k=1) as an example, the equilibrium position of the taper sleeve measured in step (2) in the simulation is Since there is no historical data, take Substituting into the controller expression of formula (10), it is calculated that the point at which the drogue sleeve is aimed is Under the interference of the head wave effect, the result of this docking is that the terminal position of the drogue sleeve is The end position of the cone is The docking error is ΔR ^(k) (T ^(k) )=0.42> _RC , so this docking is judged as failure. Then follow the same process to start the following docking attempts.

Step 3: Iterative learning controller algorithm implementation.

The form of the iterative learning control algorithm is shown in formula (10) to formula (12), where the parameters of the iterative learning controller are selected as follows

Analysis of simulation results.

Figure 3 shows the curves of the first four docking attempts in the simulation experiment. It can be seen that the first docking failed when the docking error reached 0.5m. After learning the second docking error was greatly reduced to 0.23m, it still failed. Enter the set threshold, and then the third time the docking error can be kept within the set area, the docking is considered successful. It can be seen that the algorithm is stable, and successful docking can be achieved after two docking attempts, and has a high convergence speed, so the present invention is feasible.

Claims (4)

1. An anti-interference autonomous docking control method based on terminal iterative learning control; in the air docking control problem, a controlled object is a taper sleeve on an oil receiver, a docking target is the taper sleeve at the tail end of a flexible pipe of the oil receiver, and the aim of ensuring that a taper pipe is smoothly inserted into the center of the taper sleeve under the condition of interference through iterative learning control is achieved;

one docking control problem is: p is a radical of_pr＝[x_pry_prz_pr]^TA position vector representing the controlled object, p_dr＝[x_dry_drz_dr]^TThe position vectors of the butt joint targets are uniformly defined in a certain coordinate system; the position error of the controlled object and the docking target at any time t is defined as

Wherein, Δ x_dr/pr(t),Δy_dr/pr(t),Δz_dr/pr(t) respectively represent the vectors Δ p_dr/prThree components of (t); repulsive force interference is usually at Δ p_dr/pr(t) is a function of the independent variable, and the closer the distance the greater the repulsion, eventually leading to failure of the docking; the radial error Δ R is an important index for determining whether the docking is successful or not, and is defined as a norm of a position error component in the y-axis and z-axis directions

When the successful docking requires that the controlled object reaches the terminal time T of the docking target plane, the radial error Δ R of the positions of the controlled object and the terminal time T is as small as possible, and the mathematical description is as follows: the terminal time T is represented as the first time the controlled object passes through the docking target plane Deltax in the x direction_dr/prTime of 0

T＝min_t{Δx_dr/pr(t)＝0} (3)

The radial error at the end of the time Δ R (T) is within a predetermined tolerance threshold R_CThis docking is considered successful and is defined mathematically as follows

ΔR(T)＜R_C(4)

The above formula (4) is called as a docking success rate judgment criterion, and R_CIs a constant established according to actual requirements and is also called an error threshold;

in an ideal situation without external interference, the position of the docking target will eventually stabilize at an equilibrium point, denoted asConsidering the existence of random disturbance and repulsive force disturbance, the method can be used in practical situationThe docking target will deviate from the desired equilibrium point position, i.e. the actual position p of the taper sleeve at the terminal time T_dr(T) satisfies the following relationship

Wherein,representing the position deviation due to random disturbances, w_bowIndicating a positional deviation caused by the repulsive force; repulsive force position deviation w_bowDifference Δ p between both positions_dr/prThe correlation, namely the position of the docking target can generate deviation along with the approach of the controlled object,

random position deviation w_drAre ubiquitous in nature and are generated by air, fluid or ground vibrations; due to w_drThe randomness of the control cannot be overcome by the control method, so that the control of the butt joint should be avoided at w_drThe docking control is implemented in a large number of cases; when the weather state is very poor and the taper sleeve swings randomly in the air to a large extent, the butt joint task can be almost impossible to complete and is very dangerous, so that the air butt joint is avoided in severe weather; therefore, in the actual docking control task, the random disturbance amplitude of the docking target is limited, and it is considered that the docking control is possible when the following constraints are satisfied

|w_dr|≤0.5R_C(6)

In the iterative learning process, position and error information under different iteration times need to be represented; the numerical value of a variable at the kth docking time is represented by a right superscript (k);represents the position error, T, at the time of the kth docking process^(k)Denotes the terminal time, Δ R, of the kth docking procedure^(k)(T^(k)) Represents the radial error at the k-th docking; assuming that a total of N docking attempts were made, with m successful docks, the success rate of the dock is expressed as

The radial butt joint error delta R is made by an iterative learning control method^(k)(T^(k)) With the increase and the decrease of the iteration times k, the butt joint control with high success rate is finally realized under the condition that the repulsion interference exists;

wherein A represents a controlled object, B represents an inner ring controller, and C represents iterative learning control; the iterative learning controller is a track p of a controlled object_prTrajectory p to docking target_drGenerating a suitable tracking position signal for input through iterative learning of historical dataThe function of the inner loop controller is to feed back the state vector x of the controlled object_prRealize self stable control and track given reference position signalRealizing stable fixed-point tracking and outputting a control vector u of a controlled object_pr；

The method is characterized by comprising the following steps:

step 1: inner loop controller design ensures location tracking

The method for controlling the attitude and the position is mature at present, and after the design of the inner ring controller is finished, under the ideal condition of no interference, any given coordinate system F_TThree-dimensional vector of target position ofThe controlled object can reach the designated target position at the terminal time TNamely, it isSince a tracking error actually occurs due to the presence of various kinds of interference, the position of the terminal time is expressed as follows

Wherein,tracking error vector of controlled object generated by disturbance; to meet the docking requirements, it is proposed to at least meet

|w_pr|≤0.5R_C(9)

Tracking error vector w_prThe size of the interference depends on the strength of the maneuvering performance of the airplane and the external interference; under severe weather conditions, position fluctuation is large, or the maneuvering performance of the controlled object is poor, so that the tracking accuracy is poor, and it is unrealistic to realize accurate docking control; therefore, the equation (9) is a constraint of the performance of the controlled object and the external interference, and is a basic condition for ensuring that the docking control can be successfully performed;

step 2: determining implementation of an iterative docking procedure

The application of iterative learning control requires making a scheme for the whole iterative learning process, determining when to start docking, when to finish docking and returning to where to perform the next iteration;

the iterative learning process is described below;

(1) the controlled object moves to a certain initial position at a certain distance behind the butt joint target, and the position fluctuation condition of the butt joint target and the controlled object is observed; if the external condition is good, and the random position deviation of the controlled object meets the constraint formula (6) and the constraint formula (9), the condition of successful butt joint is considered to exist;

(2) staying for a period of time, and averaging the trajectory data of the target by using the period of time to obtain the equilibrium position of the controlled object

(3) The controlled object slowly approaches the target position under the action of the inner ring controllerUntil reaching the end time T^(k)Then, the initial position is immediately returned;

(4) recording the position of the controlled object at the moment of the terminal of the current butt jointAnd docking target positionCalculating the next target position by iterative learning algorithmThen repeating the step (1) to perform the next docking attempt;

and step 3: iterative learning controller algorithm implementation

Iterative learning algorithm and controlled object positionAnd docking target locationAs input, the desired target docking position is then outputThe specific expression is as follows

WhereinFor estimating positional deviation of a docking target due to repulsive force interference, whichThe update law is as follows

WhileThe tracking deviation generated by insufficient maneuvering performance of the controlled object is counteracted by the following update law

The vector relating to two constants in the two formulae is defined as follows

Wherein,parameter(s)k_liCloser to 1 indicates higher utilization of the historical data, but slower iteration speed.

2. The anti-interference autonomous docking control method based on terminal iterative learning control according to claim 1, characterized in that: in step 1, the attitude and position control method includes a root trajectory, pole allocation and an LQR design method.

3. The anti-interference autonomous docking control method based on terminal iterative learning control according to claim 1, characterized in that: in step 2, the controlled object moves to the rear of the butt joint target by a certain distance which is more than 2 times of the length of the controlled object.

4. The anti-interference autonomous docking control method based on terminal iterative learning control according to claim 1, characterized in that: the residence time is greater than 10 s.