Double-propeller unmanned ship model identification and evaluation method based on sparse nonlinear dynamics identification

Abstract

The invention relates to the field of double-propeller unmanned ship model predictive control, in particular to a sparse nonlinear dynamics identification-based double-propeller unmanned ship model identification and evaluation method, which comprises the following steps of: combining a three-degree-of-freedom control model with characteristics of a double-propeller unmanned ship propeller to establish a double-propeller unmanned ship model; the method comprises the following steps: building a double-propeller unmanned ship model identification device, and synchronously acquiring a control input instruction and state data of an unmanned ship by autonomously controlling the unmanned ship to travel in a Z-shaped track motion; establishing a candidate function library of the double-propeller unmanned ship model, and solving to-be-identified parameters of the double-propeller unmanned ship model by adopting a sequential threshold least square method; and substituting the to-be-identified parameters of the double-propeller unmanned ship model into the double-propeller unmanned ship model to carry out Z type and rotation experiment verification, and quantitatively evaluating the precision of the double-propeller unmanned ship model through a root mean square error and a correlation coefficient. According to the method, the precision and efficiency of dynamic modeling can be remarkably improved, and reliable technical support is provided for development of an autonomous control system of the unmanned ship driven by the double propellers.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of twin-propeller unmanned ship model predictive control, and in particular to a twin-propeller unmanned ship model identification and evaluation method based on sparse nonlinear dynamics identification. BACKGROUND

[0002] With the rapid development of intelligent control technology, twin-propeller driven unmanned ships are widely used in water quality sampling, water surface rescue, water area cleaning and patrol monitoring, etc. The control system of such unmanned ships essentially belongs to a multi-input multi-output nonlinear system, and its control performance is highly dependent on the accuracy of the dynamic model. Chinese patent CN119861550A discloses a ship path tracking method based on adaptive line-of-sight guidance and fuzzy adaptive PID, including the following steps: first, preset the path point and obtain the real-time coordinates; calculate the actual desired heading angle and the ship's observed heading angle, subtract them to get the heading deviation, and calculate the heading deviation rate; input the heading deviation and the heading deviation rate into the fuzzy adaptive PID motion control module, and output the gain parameter by the fuzzy control algorithm; input the gain parameter into the PID controller to calculate the ship rudder angle; update the motion state by inputting the motion mathematical model, and form a closed loop to update the motion state to make the ship approach the preset path point; after updating the state information, the ship path is judged. However, this method is difficult to effectively handle the system coupling characteristics, and cannot consider the state and input constraints of the unmanned ship. In contrast, nonlinear model predictive control can naturally incorporate system constraints, but its control effect is directly limited by the accuracy of the underlying model.

[0003] The core challenge in ship dynamics modeling is the accurate identification of hydrodynamic parameters. Another Chinese patent CN110647041A discloses a method for accurately identifying all coefficients of an unmanned ship model, which uses an unmanned ship model all-coefficient accurate identification structure for identification. The structure includes a filter module, an integral filter module, a parameter online estimation module, and a ship model. However, the traditional all-coefficient identification method used by this invention requires a large number of towing tests and rotating arm tests to measure parameters such as added mass and damping coefficient, which is costly and time-consuming. For small unmanned ships, this mechanism-based parameter identification method has significant limitations in terms of economy and feasibility.

[0004] Another Chinese patent CN116702320A discloses a method for identifying the parameters of an unmanned ship response model based on an improved particle swarm algorithm. Although the improved particle swarm algorithm used by this invention avoids reliance on prior knowledge, its genetic operation mechanism leads to an exponential increase in computational complexity with the dimension of the parameters. Chinese patent CN117951817A discloses a method, device, equipment and medium for identifying the dynamics model of an unmanned ship. Although this invention improves the computational efficiency through gradient descent, it does not introduce sparse constraints, which can easily lead to overfitting under noise interference. SUMMARY

[0005] In order to solve the problems of strong redundancy of water power parameter calculation and high complexity of data-driven identification model in the prior art, the application provides a double-propeller unmanned ship model identification method based on sparse nonlinear dynamics identification, realizes high-precision and high-efficiency dynamics parameter identification of the unmanned ship through the construction of a technical closed loop of "modeling-data acquisition-parameter identification-verification and evaluation", and is especially suitable for the rapid dynamics modeling demand of small scientific research unmanned ships, and has significant advantages in application scenes such as water quality monitoring, water patrol and garbage cleaning. By providing a high-precision and low-cost model identification scheme, the application has promoted the process of unmanned ship control technology from the laboratory to practical application.

[0006] The application provides a double-propeller unmanned ship model identification and evaluation method based on sparse nonlinear dynamics identification, and the method comprises the following steps: The motion characteristics and the hull characteristics of the double-propeller unmanned ship are labeled based on the relative position relationship between the inertial coordinate system and the hull coordinate system, a three-degree-of-freedom maneuvering model is established, and the three-degree-of-freedom maneuvering model is combined with the propeller characteristics of the double-propeller unmanned ship to establish a double-propeller unmanned ship model; A double-propeller unmanned ship model identification device is built, the unmanned ship is controlled to run a Z-shaped trajectory motion, and the control input instructions and the state data of the unmanned ship are synchronously collected and taken as the basis for solving the to-be-identified parameters of the double-propeller unmanned ship model; A double-propeller unmanned ship model candidate function library containing a second-order polynomial term is established based on a sparse nonlinear dynamics identification framework, and a sequential threshold least square method is used to solve the to-be-identified parameters of the double-propeller unmanned ship model; The to-be-identified parameters of the double-propeller unmanned ship model are substituted into the double-propeller unmanned ship model to perform Z-shaped and circular motion verification, and the model precision is quantitatively evaluated through the root mean square error and the correlation quantity.

[0007] Further, the process of establishing the double-propeller unmanned ship model comprises: The north-east ground coordinate system of a standard ship is taken as the inertial coordinate system , the relative position relationship of the hull coordinate system of the double-propeller unmanned ship is constructed , the motion characteristics and the hull characteristics of the double-propeller unmanned ship are labeled; The center of the unmanned ship and the center of gravity coincide, wind disturbance, wave disturbance and flow disturbance are ignored, and a three-degree-of-freedom maneuvering model of the double-propeller unmanned ship is established:

[0008] wherein represents a velocity vector, u, v and r respectively represent the longitudinal velocity, the lateral velocity and the yaw rate in the body coordinate system, represents a position vector, Represents the heading angle in the inertial coordinate system. The rotation moment represents the rotational transformation matrix that transforms the velocity vector from the system frame to the inertial frame. Represents the total mass matrix. Represents the Coriolis centripetal force matrix. Represents the damping matrix. express , directional thrust and Directional torque.

[0009] The propeller's power output is controlled by adjusting the pulse width of the pulse width modulation (PWM) signal. The PWM signal vector controlling the left or right thrust is defined as follows: By combining the three-degree-of-freedom maneuvering model with the characteristics of a dual-propeller unmanned surface vessel (USV), the following dual-propeller USV model is established:

[0010] in The derivative of the three-degree-of-freedom velocity state vector is represented. This indicates the longitudinal velocity of the dual-propeller unmanned surface vessel. This indicates the lateral velocity of the dual-propeller unmanned surface vessel. This represents the yaw rate of the dual-propeller unmanned surface vessel. Indicates the longitudinal velocity of the dual-propeller unmanned surface vessel. Rate of change in direction This indicates the lateral speed of the dual-propeller unmanned surface vessel. Rate of change in direction This indicates the yaw rate of the dual-propeller unmanned surface vessel. Rate of change in direction This indicates the PWM control input signal for the port thruster. This indicates the PWM control input signal for the starboard thruster. The 15 sets of parameters to be identified represent the hydrodynamic characteristics and overall response relationship of the propulsion system of the dual-propeller unmanned surface vessel model.

[0011] Furthermore, the dual-propeller unmanned vessel model identification device includes an experimental scene, a perception module, and a remote computing processing platform.

[0012] Furthermore, in open experimental water environments such as indoor pools, a motion capture and positioning system is deployed to form a dual-propeller unmanned surface vessel (USV) model identification device. This device autonomously controls the USV to move in a Z-shaped trajectory while simultaneously collecting the USV's status data and control inputs.

[0013] Furthermore, the candidate function library for the dual-propeller unmanned surface vessel model is represented as follows:

[0014]

[0015] in The derivative of the three-degree-of-freedom velocity state vector is represented. This represents the constant coefficient matrix consisting of the parameters to be identified in the dual-propeller unmanned surface vessel model. The 15 sets of parameters to be identified represent the dual-propeller unmanned surface vessel model. This represents a candidate function library for dual-propeller unmanned surface vessel models. Indicates the three-degree-of-freedom velocity state. This represents the PWM signal vector that controls whether the signal is pushed left or right. This indicates the longitudinal velocity of the dual-propeller unmanned surface vessel. This indicates the lateral velocity of the dual-propeller unmanned surface vessel. This represents the yaw rate of the dual-propeller unmanned surface vessel. Indicates the longitudinal velocity of the dual-propeller unmanned surface vessel. and lateral speed The cross product term, Indicates the lateral speed of the dual-propeller unmanned surface vessel. and yaw rate The cross product term, Indicates the longitudinal velocity of the dual-propeller unmanned surface vessel. and yaw rate The cross product term, This indicates the PWM control input signal for the port side thruster of the dual-propeller unmanned surface vessel. This indicates the PWM control input signal for the starboard propeller of the dual-propeller unmanned surface vessel.

[0016] Furthermore, the parameters to be identified for the dual-propeller unmanned surface vessel model are solved using the sequential threshold least squares method, including: Using least squares sparse regression with L1 norm:

[0017] in The first constant coefficient matrix representing the parameters to be identified in the dual-propeller unmanned surface vessel model is represented by the matrix of constant coefficients. Row coefficient, Denotes the L1 norm used for sparsity constraints. The square of the L2 norm is the square of the Euclidean distance. The constant coefficient matrix representing the parameters to be identified in the dual-propeller unmanned surface vessel model is shown in the figure. The estimated value of the row coefficient, Indicates the first Time series data of the actual measured rate of change of velocity state of each element. This represents a candidate function library for dual-propeller unmanned surface vessel models. Indicates the three-degree-of-freedom velocity state. a vector of PWM signals representing control of left or right thrust, a tuning parameter promoting sparsity.

[0018] Further, the constant matrix is a sparse matrix with some elements being zero, so the parameter identification can adopt the least square sparse regression with L1 norm:

[0019] is the kth row coefficient of determines the dynamic expression form of the kth element; is the kth row coefficient of is the kth row coefficient of is the kth row coefficient of is the estimation value of the kth row coefficient of is the actual measured rate of change time series data of the kth element; is the corresponding actual measured change amount time series data; λ is a tuning parameter promoting sparsity. For sparse nonlinear dynamics identification, the algorithm of sequential threshold least squares is used for L1 regularization processing to obtain the model parameters.

[0020] Further, the to-be-identified parameters of the twin-propeller unmanned ship model are substituted into the nonlinear dynamics equation for verification, and the recorded data are analyzed and processed using Z-type and spiral experiments, and the performance of the unmanned ship in different test scenarios is compared.

[0021] Further, the following methods are used to quantify and evaluate the accuracy of the twin-propeller unmanned ship model through root mean square error and correlation coefficient: calculate the root mean square of the Euclidean distance between the actual position and the reference position of the twin-propelled unmanned ship model at each time :

[0022] wherein represents the actual motion state data of the twin-propelled unmanned ship at time t, represents the state data predicted by the twin-propelled unmanned ship model at time t, represents the total number of controls of the model predictive control algorithm during the experiment; calculate the correlation coefficient for quantifying and evaluating the consistency of the dynamic characteristics of the twin-propelled unmanned ship dynamics model :

[0023] wherein represents the estimation data of the motion state of the twin-propelled unmanned ship at time t.​ When or , an automatic parameter optimization cycle is triggered, the candidate function library is readjusted and the adjustment parameters are promoted to sparsity until and , wherein represents the root mean square of the Euclidean distance between the actual position and the reference position of the twin propeller unmanned ship model at each time , a standard threshold value of represents the correlation coefficient for quantitatively evaluating the consistency of the dynamic characteristics of the twin propeller unmanned ship dynamics model , a standard threshold value of

[0024] Further, the twin propeller unmanned ship model identification evaluation process of the present application adopts a closed-loop iterative optimization design, mainly including four key stages: first, a test scene including an open experimental water area, a twin propeller unmanned ship, a ground station and a perception module is built; then, a Z-shaped autonomous control experiment is performed, and the state and control input data of the unmanned ship are recorded in real time until the predetermined trajectory is completed; then, the collected data are preprocessed by Savitzky-Golay filtering, and 15 groups of key model parameters are extracted by applying a sparse nonlinear dynamics identification method; finally, the parameter accuracy is verified by a Z-shaped turning experiment, forming a closed-loop optimization mechanism of "modeling-data acquisition-parameter identification-verification evaluation". The innovation of this process lies in: the Z-shaped trajectory is used as an excitation signal, which effectively excites the three-degree-of-freedom coupling characteristics of the unmanned ship, including sway, surge and yaw; the sequential threshold least squares method is used to realize parameter sparsification selection, which reduces the complexity while ensuring the accuracy of the model; an iterative verification mechanism based on experimental data is established, and when the turning experiment result does not meet the requirements, a new round of data acquisition and parameter optimization is automatically triggered.

[0025] Compared with the prior art, the present application has the following beneficial effects: The present application innovatively introduces a sparse nonlinear dynamics identification method into the field of unmanned ship dynamics identification, and realizes automatic simplification of the model structure through L1 regularization constraint. Compared with the traditional particle swarm algorithm which only identifies 5 groups of parameters of the response model of the unmanned ship, the present method can identify 15 groups of parameters of the dynamics model of the unmanned ship.

[0026] In terms of engineering implementation, the present application uses a Savitzky-Golay filter for data preprocessing, effectively solving the noise amplification problem caused by numerical differentiation. This method can reduce the state derivative estimation error, laying a data foundation for high-precision parameter identification.

[0027] Aiming at the nonlinear coupling characteristics of the double-paddle unmanned ship, the application designs a second-order candidate function library containing Coriolis force, damping force and other key physical items, which not only retains the physical interpretability, but also avoids the overfitting problem common in traditional methods by introducing sparse constraints, and still maintains stable identification under unknown noise interference.

[0028] The "modeling-data acquisition-parameter identification-verification evaluation" system framework proposed by the application innovatively combines model identification with control system development, realizes the whole-process optimization from parameter identification to control performance evaluation verification through a closed-loop verification mechanism, and shortens the development cycle compared with the prior art.

[0029] Based on the above innovation, the application is particularly suitable for the rapid dynamic modeling needs of small scientific unmanned ships, and has significant advantages in water quality monitoring, water patrol, garbage cleaning and other application scenarios. The application provides a high-precision, low-cost model identification scheme, which effectively promotes the process of unmanned ship control technology from the laboratory to practical application. BRIEF DESCRIPTION OF DRAWINGS

[0030] In order to more clearly illustrate the technical solutions in the embodiments of the application or the prior art, the drawings needed to be used in the embodiments or prior art description will be briefly introduced as follows. Obviously, the drawings in the following description are some embodiments of the application, and other drawings can also be obtained by those skilled in the art without creative labor.

[0031] Figure 1 A flow chart of a double-paddle unmanned ship model identification and evaluation method for sparse nonlinear dynamic identification according to the application.

[0032] Figure 2 A double-paddle unmanned ship coordinate system modeling for a double-paddle unmanned ship model identification and evaluation method for sparse nonlinear dynamic identification according to the application.

[0033] Figure 3 A double-paddle unmanned ship model identification device for a double-paddle unmanned ship model identification and evaluation method for sparse nonlinear dynamic identification according to the application.

[0034] Figure 4 A double-paddle unmanned ship identification schematic diagram for a double-paddle unmanned ship model identification and evaluation method for sparse nonlinear dynamic identification according to the application.

[0035] Figure 5 A double-paddle unmanned ship model identification and evaluation process for a double-paddle unmanned ship model identification and evaluation method for sparse nonlinear dynamic identification according to the application. DETAILED DESCRIPTION

[0036] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0037] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0038] This invention provides the following technical solutions: like Figure 1 The method for identifying and evaluating a dual-propeller unmanned surface vessel model based on sparse nonlinear dynamics, as shown, specifically includes the following steps: S1: Based on the relative positional relationship between the inertial coordinate system and the hull coordinate system, mark the motion characteristics and hull characteristics of the dual-propeller unmanned vessel, establish a three-degree-of-freedom control model, and combine the three-degree-of-freedom control model with the propulsion characteristics of the dual-propeller unmanned vessel to establish a dual-propeller unmanned vessel model.

[0039] In one embodiment, such as Figure 2 As shown, the process of establishing the dual-propeller unmanned surface vessel model includes: Using the northeast coordinate system of a standard ship as the inertial coordinate system Constructing the hull coordinate system of the dual-propeller unmanned surface vessel Relative positional relationships are used to label the motion characteristics and hull features of the dual-propeller unmanned surface vessel; The double-paddle unmanned ship model established by the application is based on the following key assumptions: first, the ship body mass distribution is uniform, so that the center of the double-paddle unmanned ship is completely coincident with the center of gravity, which can effectively eliminate the additional torque item caused by uneven mass distribution, and make the dynamic equation more concise; second, the influence of environmental interference factors such as wind, wave and flow on the ship body movement is not considered in the model construction process, that is, the experiment is assumed to be carried out in an ideal static water environment. Based on these reasonable assumptions, the steering motion mathematical model of the double-paddle unmanned ship can be simplified as a rigid body dynamics system with three degrees of freedom (surge, sway and yaw), so as to establish a three-degree-of-freedom steering model of the double-paddle unmanned ship:

[0040] wherein represents a velocity vector, u, v and r respectively represent the longitudinal velocity, lateral velocity and bow angle velocity in the body coordinate system, represents a position vector, represents the bow angle in the inertial coordinate system, represents the rotation transformation matrix of the velocity vector from the body system to the inertial system, represents the total mass matrix, represents the Coriolis centripetal force matrix, represents the damping matrix, represents , directional thrust and directional moment.

[0041] The power output of the propeller is controlled by adjusting the pulse width of the pulse width modulation (PWM) signal. The PWM signal vector for controlling the left or right push is defined as The three-degree-of-freedom steering model is combined with the propeller characteristics of the double-paddle unmanned ship to establish a double-paddle unmanned ship model as follows:

[0042] wherein represents the derivative of the three-degree-of-freedom velocity state vector, represents the longitudinal velocity of the double-paddle unmanned ship, represents the lateral velocity of the double-paddle unmanned ship, represents the yaw angle velocity of the double-paddle unmanned ship, represents the rate of change of the double-paddle unmanned ship in the longitudinal velocity direction, represents the rate of change of the double-paddle unmanned ship in the lateral velocity direction, represents the rate of change of the double-paddle unmanned ship in the yaw angle velocity direction, represents the PWM control input signal of the left side propeller, This indicates the PWM control input signal for the starboard thruster. The 15 sets of parameters to be identified represent the hydrodynamic characteristics and overall response relationship of the propulsion system of the dual-propeller unmanned surface vessel model.

[0043] S2: Build a dual-propeller unmanned vessel model identification device. By autonomously controlling the unmanned vessel to move in a Z-shaped trajectory, the device simultaneously collects the control input commands and status data of the unmanned vessel, and uses these as the basis for solving the identification parameters of the dual-propeller unmanned vessel model.

[0044] In one embodiment, in an open experimental aquatic environment such as an indoor pool, a motion capture and positioning system is deployed to form a dual-propeller unmanned surface vessel (USV) model identification device. The USV autonomously controls the USV to move in a Z-shaped trajectory and simultaneously collects the USV's status data and control inputs.

[0045] In one embodiment, such as Figure 3 As shown, the constructed dual-propeller unmanned surface vessel (USV) model identification device comprises an experimental scenario, a perception module, and a remote computing platform. The experimental scenario can be an indoor experimental pool, an outdoor lake, river, or nearshore open water area. For indoor scenarios, 10 high-precision motion capture cameras (1-10) are evenly distributed around the perimeter to form a 360-degree coverage positioning network. For outdoor scenarios, GPS positioning is used. The USV and ground station interact via a router or P900 module to exchange hull status and control information, forming the perception module of the identification device. A remote computing platform for the USV is located on the side of the experimental scenario; this platform is the ground station, and its computer contains the identification software. The system's data processing hub consists of a main control computer, a router, and a data recording module, interacting with the USV in real-time via a wireless communication interface. The identification device includes a controller and a communication interface, programmed to implement autonomous yaw and Z-shaped model predictive control programs. The communication interface includes common data exchange protocols such as serial port / CAN / SBUS / TCP / UDP, providing identification and verification methods for model recognition. The ingenious feature of this unmanned surface vessel (USV) model recognition device is that the motion capture cameras are arranged in a circular array, ensuring that the USV is always within the common field of view of at least three cameras when performing Z-shaped trajectory and turning experiments, effectively solving the occlusion problem of traditional monocular vision systems.

[0046] S3: Based on the sparse nonlinear dynamics identification framework, a candidate function library for dual-propeller unmanned surface vessel models containing second-order polynomial terms is established, and the parameters to be identified for the dual-propeller unmanned surface vessel model are solved by the sequential threshold least squares method.

[0047] In one embodiment, the candidate function library for the dual-propeller unmanned surface vessel model is represented as follows:

[0048]

[0049] in The derivative of the three-degree-of-freedom velocity state vector is represented. This represents the constant coefficient matrix consisting of the parameters to be identified in the dual-propeller unmanned surface vessel model. The 15 sets of parameters to be identified represent the dual-propeller unmanned surface vessel model. This represents a candidate function library for dual-propeller unmanned surface vessel models. Indicates the three-degree-of-freedom velocity state. This represents the PWM signal vector that controls whether the signal is pushed left or right. This indicates the longitudinal velocity of the dual-propeller unmanned surface vessel. This indicates the lateral velocity of the dual-propeller unmanned surface vessel. This represents the yaw rate of the dual-propeller unmanned surface vessel. Indicates the longitudinal velocity of the dual-propeller unmanned surface vessel. and lateral speed The cross product term, Indicates the lateral speed of the dual-propeller unmanned surface vessel. and yaw rate The cross product term, Indicates the longitudinal velocity of the dual-propeller unmanned surface vessel. and yaw rate The cross product term, This indicates the PWM control input signal for the port side thruster of the dual-propeller unmanned surface vessel. This indicates the PWM control input signal for the starboard propeller of the dual-propeller unmanned surface vessel.

[0050] In one embodiment, the parameters to be identified for the dual-propeller unmanned surface vessel model are solved using the sequential threshold least squares method, including: Using least squares sparse regression with L1 norm:

[0051] in The first constant coefficient matrix representing the parameters to be identified in the dual-propeller unmanned surface vessel model is represented by the matrix of constant coefficients. Row coefficient, Denotes the L1 norm used for sparsity constraints. The square of the L2 norm is the square of the Euclidean distance. The constant coefficient matrix representing the parameters to be identified in the dual-propeller unmanned surface vessel model is shown in the figure. The estimated value of the row coefficient, Indicates the first Time series data of the actual measured rate of change of velocity state of each element. This represents a candidate function library for dual-propeller unmanned surface vessel models. Indicates the three-degree-of-freedom velocity state. This represents the PWM signal vector that controls whether the signal is pushed left or right. This represents the adjustment parameter that promotes sparsity.

[0052] In one embodiment, the constant coefficient matrix The middle part is 0, which is a sparse matrix. Therefore, parameter identification can be performed using least squares sparse regression with L1 norm.

[0053] for The coefficients in the k-th row determine... The dynamic expression of the k-th element; for for The estimated value of the coefficients in the k-th row; The time series data of the actual measured rate of change of the kth element; The parameters are: λ represents the actual measured time series data of the changes; λ is a parameter to promote sparsity. For sparse nonlinear dynamics identification, the L1 regularization process is performed using the sequential threshold least squares algorithm to obtain the model parameters.

[0054] In one embodiment, the parameters of the dual-propeller unmanned surface vessel model are shown in the table below: Table 1. Parameters to be identified for the dual-propeller unmanned surface vessel model

[0055] S4: Substitute the parameters to be identified from the dual-propeller unmanned vessel model into the dual-propeller unmanned vessel model for Z-shaped and gyroscopic experiments for verification, and evaluate the accuracy of the dual-propeller unmanned vessel model by quantitatively using root mean square error and correlation coefficient.

[0056] In one embodiment, the following method is used to quantitatively evaluate the accuracy of the dual-propeller unmanned surface vessel model using root mean square error and correlation coefficient: Calculate the root mean square distance between the actual position and the reference position of the dual-propeller unmanned surface vessel model at each moment. :

[0057] in, Indicates a dual-propeller unmanned surface vessel Real-time motion state data Indicates a dual-propeller unmanned surface vessel The state data predicted by the dual-propeller unmanned surface vessel model at any given time. This represents the total number of control operations performed by the model predictive control algorithm during the experiment. Calculate the correlation coefficients used to quantitatively evaluate the consistency of dynamic characteristics of the dual-propeller unmanned surface vessel dynamics model. :

[0058] wherein represents a twin propeller unmanned ship the estimation data of the moment motion state; when or the parameter optimization cycle is automatically triggered, the candidate function library is readjusted and the adjustment parameter promoting sparsity is adjusted until the following condition is met and wherein represents the root mean square of the Euclidean distance between the actual position and the reference position of the twin propeller unmanned ship model at each moment the standard threshold of represents the correlation coefficient for quantitatively evaluating the consistency of the dynamic characteristics of the twin propeller unmanned ship dynamics model the standard threshold of

[0059] The present application proposes a twin propeller unmanned ship model identification method based on sparse nonlinear dynamics identification, which can significantly improve the accuracy and efficiency of dynamics modeling, and provides reliable technical support for the development of autonomous control system of twin propeller driven unmanned ship.

[0060] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present application, and not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement to part or all of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for model identification evaluation of a twin-propeller unmanned ship with sparse nonlinear dynamics identification, characterized in that, Comprising the following steps: Based on the relative position relationship between the inertial coordinate system and the hull coordinate system, the motion characteristics and hull characteristics of the twin-propeller unmanned ship are marked, a three-degree-of-freedom maneuvering model is established, the three-degree-of-freedom maneuvering model is combined with the propeller characteristics of the twin-propeller unmanned ship, and a twin-propeller unmanned ship model is established; A twin-propeller unmanned ship model identification device is built, the unmanned ship is controlled to run a Z-shaped trajectory motion, the control input command and state data of the unmanned ship are synchronously collected, and they are used as the basis for solving the to-be-identified parameters of the twin-propeller unmanned ship model; Based on the sparse nonlinear dynamics identification framework, a candidate function library of the twin-propeller unmanned ship model containing second-order polynomial terms is established, and the sequential threshold least squares method is used to solve the to-be-identified parameters of the twin-propeller unmanned ship model; The to-be-identified parameters of the twin-propeller unmanned ship model are substituted into the twin-propeller unmanned ship model for Z-shaped and circular motion experiments, and the model accuracy is quantitatively evaluated by root mean square error and correlation coefficient.

2. The method according to claim 1, wherein The process of establishing the twin-propeller unmanned ship model comprises: Taking the north-east coordinate system of the standard ship as the inertial coordinate system , constructing the ship coordinate system of the twin-propeller unmanned ship The relative position relationship marks the motion characteristics and ship features of the twin-propeller unmanned ship; The center and the center of gravity of the unmanned ship coincide, the three-degree-of-freedom maneuvering model of the twin-propeller unmanned ship is established by ignoring wind disturbance, wave disturbance and flow disturbance; The power output of the propeller is controlled by adjusting the pulse width of the pulse width modulation (PWM) signal. A PWM signal vector for controlling left or right propulsion is defined as The three-degree-of-freedom steering model is combined with the characteristics of the double-paddle unmanned ship propeller to establish a double-paddle unmanned ship model as follows: wherein denotes the three degrees of freedom velocity state vector derivative, denotes the longitudinal velocity of the twin-propeller unmanned ship, denotes the lateral velocity of the twin-propeller unmanned ship, denotes the yaw angular velocity of the twin-propeller unmanned ship, denotes the rate of change of the longitudinal velocity of the twin-propeller unmanned ship in the direction, denotes the rate of change of the lateral velocity of the twin-propeller unmanned ship in the direction, denotes the rate of change of the yaw angular velocity of the twin-propeller unmanned ship in the direction, denotes the rate of change of the yaw angular velocity of the twin-propeller unmanned ship in the direction, denotes the rate of change of the yaw angular velocity of the twin-propeller unmanned ship in the direction, denotes the rate of change of the yaw angular velocity of the twin-propeller unmanned ship in the direction, denotes the PWM control input signal of the portside thruster, denotes the PWM control input signal of the starboard thruster, denotes the 15 groups of to-be-identified parameters of the twin-propeller unmanned ship model, reflecting the hydrodynamic characteristics of the twin-propeller unmanned ship and the overall response relationship of the propulsion system.

3. The method of claim 1, wherein the method is characterized by, The twin-propeller unmanned ship model identification device comprises an experimental scene, a perception module and a remote computing processing platform.

4. The method of claim 1, wherein the method is characterized by, The twin-propeller unmanned ship model candidate function library is represented as: wherein denotes the three degrees of freedom velocity state vector derivative, denotes a constant matrix composed of the to-be-identified parameters of the twin-propeller unmanned ship model, denotes 15 groups of to-be-identified parameters of the twin-propeller unmanned ship model, denotes a candidate function library of the twin-propeller unmanned ship model, denotes a three degrees of freedom velocity state, denotes a PWM signal vector for controlling left or right propulsion, denotes a longitudinal velocity of the twin-propeller unmanned ship, denotes a lateral velocity of the twin-propeller unmanned ship, denotes a yaw angular velocity of the twin-propeller unmanned ship, denotes a cross product term of the longitudinal velocity and the lateral velocity of the twin-propeller unmanned ship, denotes a cross product term of the lateral velocity and the yaw angular velocity of the twin-propeller unmanned ship, denotes a cross product term of the longitudinal velocity and the yaw angular velocity of the twin-propeller unmanned ship, denotes a PWM control input signal of the left side propeller of the twin-propeller unmanned ship, denotes a PWM control input signal of the right side propeller of the twin-propeller unmanned ship.

5. The method of claim 1, wherein the method is a method of sparse nonlinear dynamics identification of a twin-hull unmanned surface vehicle model identification and evaluation. The sequential threshold least squares method is used to solve the to-be-identified parameters of the twin-propeller unmanned ship model, which comprises: The least square sparse regression with L1 norm is used: wherein represents the estimated value of the coefficient of the constant matrix composed of the parameters to be identified of the twin-propeller unmanned ship model, the row coefficient, represents the L1 norm for sparsity constraint, represents the square of the L2 norm, i.e., the square of the Euclidean distance, represents the estimated value of the coefficient of the constant matrix composed of the parameters to be identified of the twin-propeller unmanned ship model, the row coefficient, represents the actual measured velocity state rate of change time series data of the element, represents the candidate function library of the twin-propeller unmanned ship model, represents the three-degree-of-freedom velocity state, represents the PWM signal vector for controlling the left or right push, represents the adjustment parameter for promoting sparsity.​ 6. The method of claim 1, wherein the method is a method of sparse nonlinear dynamics identification of a twin-hull unmanned surface vehicle model identification and evaluation. The to-be-identified parameters of the twin-propeller unmanned ship model are substituted into the nonlinear dynamics equation for verification, Z-shaped and circular motion experiments are used to analyze and process the recorded data and compare the performance of the unmanned ship in different test scenes.

7. The method of claim 1, wherein the method is a method of sparse nonlinear dynamics identification of a twin-hull unmanned surface vehicle model identification and evaluation. When the root mean square error and the correlation coefficient are used to quantitatively evaluate the accuracy of the twin-propeller unmanned ship model, the following methods are used: calculating the root mean square of the Euclidean distance between the actual position and the reference position of each moment of the model of the twin-propeller unmanned ship : wherein, represents a twin propeller unmanned ship time actual motion state data, represents a twin propeller unmanned ship time twin propeller unmanned ship model predicted state data, represents the total number of controls of the model predictive control algorithm during the experiment; Computing correlation coefficients for quantitatively assessing consistency of dynamic characteristics of a two-propeller unmanned ship dynamics model : wherein representing a twin-hull unmanned ship estimated data of the moment motion state; When or , an automatic parameter optimization cycle is triggered, the candidate function library is readjusted and the adjustment parameters promoting sparsity are promoted until and are satisfied, wherein represents the root mean square of the Euclidean distance between the actual position and the reference position of the dual-propeller unmanned ship model at each time represents a standard threshold of represents a standard threshold of the correlation coefficient for quantitatively evaluating the consistency of the dynamic characteristics of the dual-propeller unmanned ship dynamics model .

Images