CN-121978952-A - Model predictive control method and system for unmanned aerial vehicle with uncertain load quality

CN121978952ACN 121978952 ACN121978952 ACN 121978952ACN-121978952-A

Abstract

The invention discloses a model prediction control method and a model prediction control system for an unmanned aerial vehicle with uncertain load quality, which are used for establishing an unmanned aerial vehicle dynamic model containing load interference and converting the model into a general discrete state space system model, constructing a plurality of model prediction controllers, wherein each model prediction controller comprises a local linear model, each model prediction controller predicts the predicted value of an observed value according to the current system state based on the corresponding local linear model, stacks the predicted values of all calculated observed values to obtain the predicted value of the system observed value, and each model prediction controller calculates the predicted value of the control quantity according to the current system state based on the corresponding local linear model and stacks all calculated predicted values of the control quantity to obtain the predicted value of the system control quantity. The method can learn online and adapt to the dynamic change of the system caused by the change of the load quality, and actively cope with uncertainty, so that the robust and accurate track tracking control is realized.

Inventors

ZHOU XU
XU BENLIAN
LU MINGLI
ZHAO CAIHONG
CONG JINLIANG
SHI JIAN

Assignees

苏州工学院

Dates

Publication Date: 20260505
Application Date: 20260209

Claims (10)

1. A model predictive control method for an unmanned aerial vehicle with uncertain load quality, comprising the steps of: s01, establishing an unmanned aerial vehicle dynamic model containing load interference, and converting the model into a general discrete state space system model; S02, constructing a plurality of model prediction controllers, wherein each model prediction controller comprises a local linear model, each model prediction controller predicts a predicted value of an observed value according to the current system state based on the corresponding local linear model, and overlaps all the predicted values of the calculated observed values to obtain the predicted value of the system observed value; And S03, each model predictive controller calculates a control quantity predictive value according to the current system state based on the corresponding local linear model, and superimposes all the calculated control quantity predictive values to obtain a system control quantity predictive value.
2. The model predictive control method for an unmanned aerial vehicle with uncertain load quality according to claim 1, wherein the establishment of an unmanned aerial vehicle dynamics model including load disturbance is: Wherein, the Representing translational displacement in the x, y and z directions respectively, Respectively represent a roll angle, a pitch angle and a yaw angle, Representing the fixed moment of inertia of the body in the x, y and z directions, 、 And Representing the relative positions of the quadrotor and the load in the x, y and z directions respectively, For the length of the cable to be a length, The distance from the center of the rotor wing to the mass center of the quadrotor unmanned aerial vehicle is represented, and the point symbol and the double point symbol respectively represent corresponding primary derivation and secondary derivation; For the input to the system, Representing the thrust forces generated by the four rotors respectively, Refers to the mass of the load and, Is the quality of the unmanned aerial vehicle.
3. The model predictive control method for a unmanned aerial vehicle with uncertain load quality according to claim 2, wherein converting to a generic discrete state space system model comprises: Discretizing an unmanned aerial vehicle dynamic model into a translation subsystem and a rotation subsystem; The translation subsystem is as follows: The rotating subsystem is as follows: Wherein, the K is a control index after model discretization, 、、 The speeds of the drone in the x, y and z directions respectively, 、、 The speeds of the unmanned aerial vehicle in the rolling angle, the pitch angle and the yaw angle directions are respectively g is a gravity acceleration constant, Representing the time interval used for discretization; , , , Representing the moment of each rotor; converting the translation subsystem and the rotation subsystem into a general discrete state space system form is as follows: Wherein, the The state of the system is indicated and, Representing the output of the system and, Representing the input of the system and, Representing the interference vector caused by the load, a representing the state matrix, B representing the control matrix, C representing the observation matrix, Representing the noise of the state of the art, Representing observed noise.
4. A model predictive control method for a unmanned aerial vehicle with uncertain load quality according to claim 3, wherein the method of obtaining the predicted value of the system observation in step S02 comprises: The predicted value of the observed value predicted by each model predictive controller is: Wherein, the In order to be in the state of the system, In order to predict the control amount, 、 Is a coefficient; The predicted values of the system observations are: Wherein W is the sum of all weights of the normalization factors, The input weight of the local linear model I, I is the number of local linear models, Is the actual control quantity.
5. The model predictive control method for a unmanned aerial vehicle with uncertain load quality according to claim 4, wherein the input weight of each local linear model i Calculated by a gaussian kernel function, the gaussian kernel function is: Wherein, the In order to input the vector(s), Is the center of the i-th local linear model kernel, The shape and size of the gaussian model are determined.
6. The model predictive control method for an unmanned aerial vehicle with uncertain load quality according to claim 5, wherein the system control amount predicted value obtained in step S03 is: Wherein, the The control quantity predicted value of the next time step of the local linear model i.
7. The model predictive control method for an unmanned aerial vehicle of claim 6, wherein the control amount predictive value calculating method comprises: optimal predicted value of next system state Obtained as follows: wherein Z is And Cross covariance of (2); Solving by a solver to obtain an optimal predicted value of the next system state Thereafter, the optimal estimate of the state and/or the estimate of the output in the entire prediction domain from time k+1 to k+N is obtained as follows: Wherein i=1,..n, N is the predicted total number of steps; by minimizing a control objective function in the following formula, a control quantity predicted value is obtained 、 : Wherein n is the number of predicted steps, As a trajectory of the moment k, S is a matrix of weights that are to be used, S is symmetrical, positive and secondary , 。
8. Model predictive control system for a unmanned aerial vehicle with uncertain load quality, characterized in that it implements a model predictive control method for a unmanned aerial vehicle with uncertain load quality according to any of claims 1 to 7, comprising: The model construction module is used for establishing an unmanned aerial vehicle dynamic model containing load interference and converting the model into a general discrete state space system model; The system observation value prediction module is used for constructing a plurality of model prediction controllers, each model prediction controller comprises a local linear model, each model prediction controller predicts the prediction value of the obtained observation value according to the current system state based on the corresponding local linear model, and all the calculated prediction values of the obtained observation values are overlapped to obtain the prediction value of the system observation value; and each model prediction controller calculates a control quantity predicted value according to the current system state based on the corresponding local linear model, and superimposes all calculated control quantity predicted values to obtain the system control quantity predicted value.
9. A computer storage medium having stored thereon a computer program, characterized in that the computer program, when executed, implements the model predictive control method for a unmanned aerial vehicle with uncertain load quality according to any of claims 1-7.
10. An electronic device comprising a memory and a processor, wherein the memory has stored therein a computer program, the processor running the computer program stored on the memory, the computer program when executed implementing the model predictive control method for unmanned aerial vehicle with uncertain load quality according to any of claims 1-7.

Description

Model predictive control method and system for unmanned aerial vehicle with uncertain load quality Technical Field The invention belongs to the technical field of unmanned aerial vehicle control, and relates to a model predictive control method and a system for an unmanned aerial vehicle with uncertain load quality. Background Unmanned Aerial Vehicle (UAV), especially four rotor unmanned aerial vehicle, because of its mechanical structure is simple, possess perpendicular take off and land and hover ability, have wide application prospect in fields such as commodity circulation, search and rescue, agriculture. However, the quadrotor unmanned itself is an under-actuated system, and only four inputs are required to control six degrees of freedom, which has challenges for stability and control itself. When carrying the suspension load, the swing of the load can introduce complicated external disturbance, so that the dynamics characteristic of the four-rotor unmanned aerial vehicle is obviously changed, and the flight stability and the track tracking performance of the four-rotor unmanned aerial vehicle are seriously affected. In practice, the quality of the load is often unknown or changes suddenly during transport, and this uncertainty translates into time-varying external disturbances. Aiming at the problem of uncertain load quality, the traditional control method is mainly dependent on the self robustness of a controller to passively cope with uncertainty, when the load quality is changed greatly or dynamic change is severe, the control performance is reduced sharply, even the system is unstable, and the model prediction control method is usually based on a fixed and pre-calibrated system model although having a certain feedforward optimization capability, when the actual system dynamics is mismatched with the model due to the load quality change, the control performance is obviously deteriorated, and the adaptive capacity to time-varying dynamics is lacking. Therefore, there is an urgent need in the art for a control method that can actively learn and adapt to task parameter changes such as load quality on line, so as to ensure stable and accurate flight of the quadrotors in an uncertainty environment. Disclosure of Invention The invention aims to provide a model predictive control method and a system for an unmanned aerial vehicle with uncertain load quality, which can learn online and adapt to system dynamic changes caused by load quality changes, and actively cope with uncertainty, so that robust and accurate track tracking control is realized. The technical solution for realizing the purpose of the invention is as follows: A model predictive control method for unmanned aerial vehicle with uncertain load quality, comprising the steps of: s01, establishing an unmanned aerial vehicle dynamic model containing load interference, and converting the model into a general discrete state space system model; S02, constructing a plurality of model prediction controllers, wherein each model prediction controller comprises a local linear model, each model prediction controller predicts a predicted value of an observed value according to the current system state based on the corresponding local linear model, and overlaps all the predicted values of the calculated observed values to obtain the predicted value of the system observed value; And S03, each model predictive controller calculates a control quantity predictive value according to the current system state based on the corresponding local linear model, and superimposes all the calculated control quantity predictive values to obtain a system control quantity predictive value. In the preferred technical scheme, the establishment of the unmanned aerial vehicle power model containing load interference is as follows: Wherein, the Representing translational displacement in the x, y and z directions respectively,Respectively represent a roll angle, a pitch angle and a yaw angle,Representing the fixed moment of inertia of the body in the x, y and z directions,、AndRepresenting the relative positions of the quadrotor and the load in the x, y and z directions respectively,For the length of the cable to be a length,The distance from the center of the rotor wing to the mass center of the quadrotor unmanned aerial vehicle is represented, and the point symbol and the double point symbol respectively represent corresponding primary derivation and secondary derivation; For the input to the system, Representing the thrust forces generated by the four rotors respectively,Refers to the mass of the load and,Is the quality of the unmanned aerial vehicle. In a preferred technical solution, the conversion into the general discrete state space system model includes: Discretizing an unmanned aerial vehicle dynamic model into a translation subsystem and a rotation subsystem; The translation subsystem is as follows: The rotating subsystem is as follows: Wherein, the K is a control index