Search

CN-121072045-B - Unmanned aerial vehicle power system and observation method based on crew filtering algorithm

CN121072045BCN 121072045 BCN121072045 BCN 121072045BCN-121072045-B

Abstract

The invention belongs to the technical field of unmanned power systems, and discloses an unmanned power system and an observation method based on an operator-gathering filtering algorithm, wherein the method comprises the steps of firstly modeling an electric power part of the unmanned power system, establishing a motor model under static coordinates, then establishing a motor state space model and an observer model, modeling the error in the observer model, limiting the error to an ellipsoid set, establishing an error system comprising noise variables, obtaining a recurrence relation of errors at adjacent moments, solving and obtaining the gain of a filter in the observer model, and realizing dynamic updating of the gain of the observer, thereby realizing high-precision observation of the unmanned aerial vehicle power system. According to the invention, the observer of the unmanned aerial vehicle power system is constructed based on the set member theory, so that the optimal observation performance is realized under the condition that model parameters are time-varying, meanwhile, the noise in the motor is modeled, the unknown but bounded noise characteristic is described by an ellipsoid set, and the observation precision in a noise environment is improved.

Inventors

  • XU LEI
  • PANG ZHI
  • LIU YOUHUI

Assignees

  • 深圳市好盈科技股份有限公司

Dates

Publication Date
20260508
Application Date
20251105

Claims (9)

  1. 1. An observation method based on a member filtering algorithm is characterized by comprising the following method steps: Modeling an electric power part of the unmanned aerial vehicle power system, and establishing a motor model under a static coordinate, wherein the motor model comprises a voltage balance equation and a back electromotive force equation; based on a motor model in a static coordinate system, a motor state space model is established, and stator current and counter electromotive force in the static coordinate system are used as state variables; Establishing an observer model based on the motor state space model, wherein the observer model estimates an observed value based on the filter gain; Discretizing a filter in an observer model, discretizing variables in the observer model, and describing the observer model by a time-varying discrete state space equation; Modeling the error in the observer model, limiting the error to an ellipsoid set, modeling the noise, limiting the noise to an ellipsoid set, establishing an error system comprising noise variables, obtaining a recursive relationship of the errors at adjacent moments, expressed as: ; Wherein, the Representing the estimated error of the current time instant, Representing the estimated error at the next moment in time, Representing a matrix of discrete systems that are, Representing a discrete matrix of observer gains, The output matrix is represented by a representation of the output matrix, The noise is represented by a characteristic of the noise, Represents an index of an integer number, The period of the sampling is indicated and, Represent the first Sampling time; The method comprises the steps of defining an optimization problem, taking the limitation of errors in an ellipsoid set as constraint and the minimization of the size of the ellipsoid set as a target, solving the optimization problem, obtaining the gain of a filter in an observer model, and realizing the observation of an unmanned power system through the observer model; the optimization problem is expressed as: ; Wherein, the Representing the error covariance matrix for the next instant.
  2. 2. The observation method based on the member-collecting filtering algorithm according to claim 1, wherein the modeling is performed on the electric power part of the unmanned aerial vehicle power system, and a motor model under the static coordinates is established, and is expressed as: ; Wherein, the And Representing the stator current in a stationary coordinate system, And Representing the voltage in the stationary coordinate system, And Is the back electromotive force in a static coordinate system and the angular velocity of the rotor In relation to the use of a liquid crystal display device, The resistance of the line is indicated and, The inductance of the wire is indicated, Indicating the electrical angle of the rotor.
  3. 3. The observation method based on the member filtering algorithm according to claim 1, wherein the motor state space model is established based on a motor model under a static coordinate system, and is expressed as: ; Wherein, the And Representing the stator current in a stationary coordinate system, And Representing the voltage in the stationary coordinate system, And Is the back electromotive force in a static coordinate system and the angular velocity of the rotor In relation to the use of a liquid crystal display device, The resistance of the line is indicated and, The inductance of the wire is indicated, And The measured value is represented by a measurement value, Representing noise.
  4. 4. The observation method based on the member filtering algorithm according to claim 1, wherein the observer model is built based on a motor state space model, expressed as: ; Wherein, the And Representing the stator current in a stationary coordinate system, And Representing the voltage in the stationary coordinate system, And Is the back emf in a stationary coordinate system, Indicating the angular velocity of the rotor, The resistance of the line is indicated and, The inductance of the wire is indicated, I.e. with digital subscripts Representing the filter gain, i= {1,2,3,4}, j= {1,2}, And Representing the current observed by the observer, And Representing the back emf observed by the observer, And Representing the measured value.
  5. 5. The observation method based on the member filtering algorithm according to claim 1, wherein discretizing is performed on a filter in an observer model, and variables in the observer model are discretized, and the observer model is described by a time-varying discrete state space equation, which is expressed as: ; Wherein, the The state vector is represented as a function of the state vector, The input matrix is represented as such, The vector of the input voltage is represented as, Represent the first The sampling instants.
  6. 6. The observation method based on the set membership filtering algorithm according to claim 1, wherein errors in the observer model are modeled, and the errors are limited to an ellipsoid set, expressed as: ; Wherein, the Representing an error covariance matrix, for defining the shape and size of the set of ellipsoids.
  7. 7. The observation method based on the member-collecting filtering algorithm according to claim 1, wherein the method comprises the steps of solving an optimization problem, obtaining the gain of a filter in an observer model, and realizing the observation of an unmanned aerial vehicle power system through the observer model, and comprises the following steps: Equivalently transforming the optimization problem into a linear matrix inequality problem, solving the gain of a filter in an observer model, and substituting the gain into the observer; Taking out the two-phase back electromotive force obtained by observation of the observer: ; the angle and the rotating speed of the motor rotor are obtained through a phase-locked loop, and are expressed as follows: obtained by trigonometric function transformation: ; Obtained by PI converter: ; obtaining an estimated angle by integrating the speed: ; Wherein, the And Representing the back emf, Indicating the angular velocity of the rotor at the present moment, The rotor angle at the present moment is indicated, And Representing the proportional gain and the integral gain of the PI converter respectively, The integer index is represented, with subscript n replacing (nN) in the discrete system.
  8. 8. The observation method based on the member filtering algorithm according to claim 1, wherein when the condition is met, the calculation and update of the filter gain of the observer model are triggered, and the trigger mechanism is expressed as: ; Wherein, the Indicating the angular velocity of the rotor at the present moment, Representing a preset adaptive rate function, Indicating that the preset threshold value is to be reached, When 1, the calculation and update of the filter gain of the observer model are triggered.
  9. 9. An unmanned aerial vehicle power system, characterized in that according to the observation method based on the crew filtering algorithm of any one of claims 1-8, the filter gain of the observer model is calculated and updated, so as to realize the motor state observation of the unmanned aerial vehicle power system.

Description

Unmanned aerial vehicle power system and observation method based on crew filtering algorithm Technical Field The invention belongs to the technical field of unmanned power systems, and particularly relates to an unmanned power system and an observation method based on an operator filtering algorithm. Background The core system model of the unmanned power suit is usually a multiple-input multiple-output dynamic system, and state variables of the model are controlled by torque and force generated by motor thrust. To accurately control these states, which cannot all be measured directly, the system generally employs a fixed gain observer that estimates and reconstructs all internal states of the system in real time based on a dynamic model of the system, using the error between a small amount of available sensor data and the model's predicted value, multiplied by a pre-calculated, fixed gain matrix. The method has high calculation efficiency and easy realization, and provides key state information for stable flight and anti-interference control of the unmanned aerial vehicle. However, because the system model of the unmanned aerial vehicle power suit is a time-varying system model, the traditional observer adopts a strategy of fixed gain, and adverse phenomena such as parameter failure and observer divergence are easy to occur under complex working conditions such as unmanned aerial vehicle flight. In order to avoid these adverse phenomena, the prior art proposes an observer with adaptive gain to realize more reliable state observation under a time-varying system model, but the existing observer model cannot effectively consider the influence of various internal noises, and has the defect of poor robustness to noise environments. Disclosure of Invention The invention aims to overcome the defect that a system model of an unmanned aerial vehicle power system in the prior art is poor in noise environment robustness, so as to provide the unmanned aerial vehicle power system and an observation method based on a member collecting filtering algorithm. An observation method based on a set member filtering algorithm comprises the following method steps: Modeling an electric power part of the unmanned aerial vehicle power system, and establishing a motor model under a static coordinate, wherein the motor model comprises a voltage balance equation and a back electromotive force equation; based on a motor model in a static coordinate system, a motor state space model is established, and stator current and counter electromotive force in the static coordinate system are used as state variables; Establishing an observer model based on the motor state space model, wherein the observer model estimates an observed value based on the filter gain; Discretizing a filter in an observer model, discretizing variables in the observer model, and describing the observer model by a time-varying discrete state space equation; Modeling the error in the observer model, limiting the error in an ellipsoid set, modeling the noise, limiting the noise in the ellipsoid set, establishing an error system comprising noise variables, and obtaining a recurrence relation of errors at adjacent moments; The method comprises the steps of defining an optimization problem, taking the limitation of errors in an ellipsoid set as constraint and the minimization of the size of the ellipsoid set as a target, solving the optimization problem, obtaining the gain of a filter in an observer model, and realizing the observation of an unmanned power system through the observer model. Further, modeling is performed on the electric power part of the unmanned aerial vehicle power system, and a motor model under a static coordinate is established, and is expressed as follows: ; Wherein, the AndRepresenting the stator current in a stationary coordinate system,AndRepresenting the voltage in the stationary coordinate system,AndIs the back electromotive force in a static coordinate system and is related to the angle of the rotorIn relation to the use of a liquid crystal display device,The resistance of the line is indicated and,Representing the line inductance. Further, based on the motor model under the static coordinate system, a motor state space model is established, which is expressed as: ; Wherein, the AndRepresenting the stator current in a stationary coordinate system,AndRepresenting the voltage in the stationary coordinate system,AndIs the back electromotive force in a static coordinate system and is related to the angle of the rotorIn relation to the use of a liquid crystal display device,The resistance of the line is indicated and,The inductance of the wire is indicated,AndThe measured value is represented by a measurement value,Representing noise. Further, an observer model is built based on the motor state space model, expressed as: ; Wherein, the AndRepresenting the stator current in a stationary coordinate system,AndRepresenting the voltage in the stationar