1. Ship motion model
Scholars at home and abroad often assume that the hull is a rigid body, and based on the rigid body motion theory to study the modeling of ship maneuvering motion, the established model is mainly
There must be
four types:
Abkowitz integral model
, MMG separation model ( mathematical modeling group model) , matrix-vector model , and response model . The Abkowitz integral model is based on the third-order Taylor series expansion to establish the function between the ship's motion state variables and the propulsion system parameters to describe the hydrodynamic forces and moments acting on the hull, propeller, rudder, and their interactions. This model has high precision in describing the ship's maneuvering motion, but the model is extremely complex and involves a large number of parameters, and some parameters cannot be identified and have no physical meaning. The MMG separate model analyzes and models the forces and moments acting on the hull, propeller, rudder, and their interactions. Each parameter in the model has physical meaning, but under the premise of ensuring sufficient accuracy, the model has many parameters and strong nonlinearity, which makes the identification of model parameters difficult. The matrix-vector model was proposed by Fossen, which describes the force and moment acting on the ship in the form of matrix vector, which is convenient for analyzing the stability and passivity of the ship, and facilitates the design of the ship motion controller. The response model describes the response relationship of the bow to the steering, and the most representative one is the Nomoto first-order / second-order linear and nonlinear model .
2. Model identification method
1. Online identification
Research on the online identification and reconfiguration control technology of typical power system faults. At the same time, based on the online evaluation technology of the carrying and control capabilities and the parameter reconfiguration control technology of the online identification of the stability margin, the online change of the target and the online planning of the trajectory can be realized
.
By applying excitation to the closed-loop system, the frequency domain identification method is used to solve the frequency characteristic function of the system from the input and output data, and the stability margin of the system is obtained
. Based on the online identification results of the stability margin, adaptive parameter tuning control is adopted.
(1) First-order linear model identification
[1]
Wu
Bo
, et al. Design of
self-adaptive discrete sliding mode controller for heading
of unmanned
boat
. China Navigation 2021
In
the case of uncertain changes in the motion response model of the unmanned boat due to the influence
of
the external
environment
, in order to achieve effective control of the heading of the unmanned boat , a modelmotion based on online identification is designed . Adaptive Discrete Sliding Mode Controller . recursive least squares
Online identification
of the
motion response model of the
unmanned vehicle . Based on the online identification model , an adaptive discrete sliding mode controller based on exponential reaching law is designed .
(2) Data-driven identification
[2] Geng Lingbo, et al.: Data-driven underwater vehicle online modeling and motion control technology. Digital Ocean and Underwater Attack and Defense 2021
Learning from the Transformer network structure widely used in the field of machine translation, some adaptive transformation is carried out on the basis of it, and the obtained
Neural network models for modeling underwater vehicles.
The encoder is composed of one-dimensional convolution and mainly receives historical data for learning the current state of the submersible. decoder by
The LSTM is composed of mainly receiving future input data, which is used to drive the submersible to reach the future state based on the historical state.
The Encoder
input is
the state quantity and manipulation control quantity of the past 10 time steps. The state quantity includes pitch angle, roll angle, acceleration and
Angular velocity and manipulation control volume include rudder angle control volume and thruster control volume.
3. Model-free track control
1. Self-interference rejection control
The algorithm uses data-driven thinking, regards all uncertain factors acting on the controlled system as unknown disturbances, estimates and compensates unknown disturbances in real time according to system input and output information [8], has a simple structure, is easy to implement, and has been successfully implemented
. It is used in practical applications in many fields such as robots
and wind farms.
[1] Li Shijie. Model-free adaptive track control for ship self-disturbance rejection. China Ship Research 2023
4. Model Reference Adaptive Control
In adaptive control, model reference adaptive control (Model Reference Adaptive Control, MRAC) is a common method. MRAC aims to match the output of the plant with the output of the reference model, and achieves this goal by adjusting the controller parameters on-line. It is based on a model reference that incorporates the desired system response characteristics. The parameters of the controller are updated by continuously observing the difference between the output of the system and the output of the reference model, and are updated through an adaptive algorithm to achieve output matching.
The key steps of model reference adaptive control include:
-
Design Reference Model: Determine the desired system response characteristics and represent them as the output of the reference model.
-
Determine the controller structure: Choose an appropriate controller structure so that the controller parameters can be adjusted in real time based on the error signal.
-
Define the error signal: Compute the difference between the output of the system and the output of the reference model as the error signal.
-
Design an adaptive algorithm: According to the error signal and adaptive rules, design the controller parameter update algorithm to minimize the error.
-
Implement the controller: connect the designed controller with the controlled object, and adjust the parameters by continuously observing the error between the system output and the reference model output.
It should be noted that model reference adaptive control requires a certain understanding of the dynamic characteristics of the controlled object, and requires system identification to obtain the system model. In addition, the convergence and stability of the controller parameters are also issues that need to be considered.
Model reference adaptive control has applications in many fields, such as aircraft control, robot control, industrial process control, etc. It can cope with the changes and uncertainties of system dynamic characteristics, and provide better control performance and robustness.
Regenerate response