A hovercraft vector thrust level distribution method based on improved golden eagle optimization algorithm

Abstract

The present application belongs to the technical field of ship thrust distribution, and particularly relates to a full-cushion hovercraft vector thrust hierarchical distribution method based on an improved golden eagle optimization algorithm. The present application solves the problem of uneven distribution of thrust of each propeller of the full-cushion hovercraft to a large extent by reasonably planning the priority distribution order of each propeller, using a hierarchical distribution strategy and a multi-optimization target method, fully considers the execution efficiency of each propeller, improves the execution efficiency and service life of the propeller, and reduces the overall fuel consumption of the hovercraft. The improved multi-target golden eagle optimization algorithm proposed by the present application can obtain the Pareto optimal solution of the loss function of the hierarchical distribution strategy within a limited number of steps, grade the multi-target function, and improve the calculation efficiency of the multi-target golden eagle optimization algorithm.

Description

Technical Field

[0001] This invention belongs to the field of ship thrust distribution technology, specifically relating to a method for allocating vector thrust hierarchy of a fully cushioned hovercraft based on an improved Golden Eagle optimization algorithm. Background Technology

[0002] A fully-cushioned hovercraft is an amphibious special vessel that uses a cushioning system to levitate in complex environments such as water, land, and swamps, and is widely used in naval equipment and civilian rescue. Thrust distribution technology, as a crucial guarantee for the safe navigation of hovercraft, has previously been studied primarily in the coordination and matching of the air rudder and propeller, but this approach has its shortcomings and incompleteness. Firstly, the rudder-propeller coordination does not take into account the use of the hovercraft's other propulsion unit, the bow nozzle. When the hovercraft is traveling at low speeds or turning, the flexibility of the bow nozzle can be used to generate lateral turning torque, improving the hovercraft's turning efficiency. Secondly, the thrust distribution fails to consider the actuation efficiency of the hovercraft's propellers, applying the same optimization order to different actuators. However, the reality is that when the hovercraft is traveling at high speeds, the air rudder's turning efficiency is very high, while at low speeds, more coordination and matching between the propeller and bow nozzle are required. Summary of the Invention

[0003] The purpose of this invention is to solve the problem that previous thrust distribution problems only involved the control of the roll and pitch angular velocities of the hovercraft, ignoring the problem that the fully cushioned hovercraft is prone to nose-down situations. The invention provides a method for vector thrust hierarchy distribution of fully cushioned hovercraft based on the improved Golden Eagle optimization algorithm.

[0004] A method for all-cushioned hovercraft vector thrust hierarchy allocation based on an improved Golden Eagle optimization algorithm includes the following steps:

[0005] Step 1: Determine the propulsion system layout of the fully cushioned hovercraft, establish the angular mathematical model of the fully cushioned hovercraft, and load the parameters of each thruster; the parameters of the thrusters include: thruster type and number, thruster thrust direction variable range, thrust magnitude variable range, thrust direction change rate range, and thrust magnitude change rate range.

[0006]

[0007] in, For air rudder torque, For propeller torque, For the bow nozzle torque, These are the moments of inertia along the x, y, and z axes, respectively. These are respectively the roll, trim, and turning angular velocity;

[0008] Step 2: Design virtual control input based on active disturbance rejection control technology

[0009]

[0010] in, For the environmental disturbances observed by the observer and the total uncertainty of the system; k p ,k q ,k r For tracking error proportional feedback gain; ψ* represents the desired angular velocity;

[0011] Step 3: Design a strategy for allocating the thruster hierarchy of the hovercraft;

[0012] The difference between the virtual control input and the actual controller output is set as the loss function Γ,k for simulating the time series:

[0013]

[0014] The propulsion constraints of a hovercraft are rudder angle constraints and thrust vector constraints, which are Ω respectively. A,k ∈Θ A,k ,Ω F,k ∈Θ F,k H u X T This is the actual control output of the hovercraft. For virtual control input;

[0015] Hierarchical allocation step 1: Since the hovercraft has the highest yaw rate, prioritizing the allocation of air rudders will yield better results. The loss function for allocating air rudders is denoted as Γ. 1,k :

[0016]

[0017] Hierarchical Allocation Step 2: Given the low rotational efficiency of the propeller and bow nozzle in Hierarchical Allocation Step 1, optimize the thrust vectors of the propeller and bow nozzle, denoted as the loss function Γ. 2,k :

[0018]

[0019] This indicates the optimal solution using the hierarchical allocation method in step 1;

[0020] Hierarchical allocation step 3: Under the premise of hierarchical allocation step 2, minimize the thrust of the propeller and the bow nozzle to reduce the energy consumption of the engine and gas turbine. Design the energy minimization function denoted as E. k :

[0021]

[0022] This indicates the optimal solution using the hierarchical allocation step 2;

[0023] Step 4: Since the efficiency of each propeller is different at different speeds of the hovercraft, design a hierarchical allocation weight coefficient to make the hierarchical allocation strategy more in line with the actual propeller allocation during the actual navigation of the hovercraft.

[0024] Step 5: Solve the hierarchical allocation step loss function using the improved multi-objective Golden Eagle optimization algorithm.

[0025] Furthermore, the fully-cushioned hovercraft in step 1 has two air propellers, each with two synchronized air rudders behind it, and a bow nozzle on each side of the ship, which can rotate 360°.

[0026]

[0027] Among them, F xR F yR For the forces exerted by a single aerodynamic rudder in the x and y directions, C Rx (δ R C Ry (δ R ) represents the aerodynamic coefficient, obtained from aerodynamic test data of the vertical air rudder; u aa The velocity of the rudder inflow caused by the propeller wake; S R ρ is the area of ​​the vertical air rudder. a air density; x R y R z R X represents the distance of the aerodynamic rudder's center of gravity along the x, y, and z directions; P This is the distance between the pressure center and the rudder shaft;

[0028]

[0029] Among them, z P y P F represents the distance between the propeller's center of gravity and the z-axis; Pi This refers to the magnitude of the propeller force;

[0030]

[0031] Where, δ Si F is the bow nozzle deflection angle. Si The magnitude of the bow nozzle force; x Si y Si z Si The distances of the bow nozzle in the x, y, and z directions; [J] x J y J z ] are the moments of inertia of the x, y, and z axes, respectively.

[0032] Furthermore, step 2 specifically includes:

[0033] The equation for angular velocity is written as Form, ψ = [p; q; r], T w =[T p ;T q ;T r The total model uncertainty and total environmental disturbances for the hovercraft's heel, trim, and yaw.

[0034] The relationship between the actual control input and the actual control input is denoted as:

[0035]

[0036] Among them, H u = [H1,H2,H3,H4,H5,H6,H7]∈R 3×7 ;

[0037] X = [δ] R ,F P1 ,F P2 ,F S1 sinδ S1 ,F S2 sinδ S2 ,F S1 cosδ S1 ,F S2 cosδ S2 ];

[0038] According to the Lagrange mean value theorem, the aerodynamic rudder torque is transformed as follows:

[0039]

[0040] in, Lagrange expansion point and endpoint δ R For any point between these points, combining the above equation with the angular velocity equation, we get:

[0041]

[0042] Combining the angular velocity equation and the propeller and bow nozzle moment equations, the lever arm is extracted to obtain:

[0043]

[0044] Where J = diag(J x J y J z );

[0045] A second-order extended state observer is designed using the bandwidth method to compensate for the nonlinear uncertainties and total environmental disturbances of the angular velocity control system. 01 ,β 02 For system bandwidth, β p ,β q ,β r The bandwidth control objectives in the p, q, and r directions are respectively to make the angular velocity track the time-varying reference signal ψ*(t)=[p*(t),q*(t),r*(t)]. T :

[0046]

[0047] Where, β 01 =2diag(β) p ,β q ,β r ), To compensate for uncertainties and environmental disturbances in the angular velocity control system, a virtual control law is designed using error proportional feedback control.

[0048]

[0049] k p ,k q ,k r This is the proportional feedback gain for tracking error.

[0050] Furthermore, step 4 specifically includes:

[0051] The weighting coefficients for step 1 of the hierarchical allocation are set as follows:

[0052]

[0053] Among them, u k v k Let K represent the lateral and longitudinal velocities of the fully-cushioned hovercraft at time k.

[0054] The weighting coefficients for step 2 of the hierarchical allocation are set as follows:

[0055]

[0056] The weight coefficient for step 3 of the hierarchical allocation is always 1.

[0057] Furthermore, step 5 specifically includes:

[0058] Step 5.1: Confirm the objective function of the multi-objective Golden Eagle algorithm as F = {{Γ} 1,k , Γ 2,k}, E k The weight coefficients of the objective function are {{η}. 1,k η2,k The cruising vector in the Golden Eagle optimization algorithm is},1}, Attack vector is Prey is the optimal variable

[0059] Initialize the number of golden eagles and estimate the loss function Γ. 1,k and Γ 2,k and the energy minimization function E k The value of the initialization function is used to initialize the Golden Eagle memory bank and the attack coefficient p. a and cruise coefficient p c Among them, the attack coefficient increases with the number of iterations, while the cruise coefficient decreases with the number of iterations, simulating the golden eagle's transition from cruise to attack, and the process of approaching the target prey from the initial position.

[0060] Step 5.2: For each iteration k, update p a and p c The value of p is calculated, and the crowding distance of existing memory bank members is calculated, causing the Golden Eagle to gradually switch from cruise mode to attack mode, and p is updated. a and p c The method is as follows:

[0061]

[0062] Where T is the total number of iterations, and k is the current iteration. and These are the initial coefficients. and Set the final target value; set the current position as... The terminal is the optimal variable. The attack model of the Golden Eagle is then represented as follows:

[0063] Step 5.3: The Golden Eagle cruises to move to a new location, described as follows:

[0064]

[0065] in, Let i be the step size of the golden eagle i in k iterations;

[0066] Combining a hierarchical allocation strategy, the loss function Γ for the new position is calculated according to a certain priority. 1,k ,Γ 2,k E k If the new position is a Pareto non-dominated solution relative to existing library members and the external library is not full, then the new solution is added to the library; otherwise, the sparse distance is calculated, and a roulette wheel with sparse distance weights is used to select the eliminated member and replace it with the new position. This process is iterated until the algorithm requirements are met.

[0067] The beneficial effects of this invention are as follows:

[0068] This invention addresses the problem of uneven thrust distribution among the thrusters of a fully-cushioned hovercraft by rationally planning the priority allocation order of each thruster and employing a hierarchical allocation strategy and multiple optimization objectives. It fully considers the execution efficiency of each thruster, improving their efficiency and lifespan, and reducing the overall fuel consumption of the hovercraft. The improved multi-objective Golden Eagle optimization algorithm proposed in this invention can obtain the Pareto optimal solution for the hierarchical allocation strategy loss function within a finite number of steps, and distinguishes the levels of the multi-objective functions, thus improving the computational efficiency of the multi-objective Golden Eagle optimization algorithm. Attached Figure Description

[0069] Figure 1 Schematic diagram of thrust vector distribution principle for a fully cushioned hovercraft.

[0070] Figure 2 This is a schematic diagram of the present invention.

[0071] Figure 3 This is a flowchart of the improved Golden Eagle optimization algorithm in this invention. Detailed Implementation

[0072] The present invention will now be further described with reference to the accompanying drawings.

[0073] The thrust vector distribution principle diagram of the fully cushioned air-cushioned vehicle is as follows: Figure 1 As shown, traditional thrust distribution optimization methods for hovercraft mainly employ pseudo-inverse algorithms, sequential quadratic optimization, and direct allocation methods. These algorithms fail to consider the priority order of thruster usage, even though the thruster efficiencies vary. Furthermore, they do not address the real-time optimization of control inputs, failing to provide optimal control inputs promptly.

[0074] The purpose of this invention is to address the problem that previous thrust distribution problems only involved the control of the roll and pitch angular velocities of hovercraft, neglecting the tendency of fully-cushioned hovercraft to pitch down. This invention aims to rationally plan the priority allocation order of each thruster, using a hierarchical allocation strategy and a multi-objective optimization method to improve thruster utilization efficiency, extend the service life of each thruster, and reduce the overall fuel consumption of the hovercraft. Furthermore, it improves the multi-objective Golden Eagle optimization algorithm by using hierarchical allocation weight coefficients to optimize the multi-objective function, thereby increasing the computational efficiency of the multi-objective Golden Eagle optimization algorithm.

[0075] A method for all-cushioned hovercraft vector thrust hierarchy allocation based on an improved Golden Eagle optimization algorithm includes:

[0076] Step 1: Determine the propulsion system layout of the fully-cushioned hovercraft, establish an angular mathematical model of the fully-cushioned hovercraft, and load the parameters of each thruster. The thruster parameters include: thruster type and number, thrust direction variable range, and thrust magnitude variable range. Establish the angular velocity control model of the hovercraft.

[0077] The fully-cushioned hovercraft of this invention has two air propellers, each with two synchronously linked air rudders, and one bow nozzle on each side of the hull, enabling full azimuth rotation. A mathematical model for the angular velocity control of the fully-cushioned hovercraft is established as follows:

[0078]

[0079] in For aerodynamic rudder torque, and Let C be the force exerted by a single aerodynamic rudder in the x and y directions. Rx (δ R C Ry (δ R ) represents the aerodynamic coefficient, obtained from aerodynamic test data of the vertical air rudder, u aa The velocity of the rudder inflow caused by the propeller wake, S R For the vertical air rudder area, ρ a x is the air density R y R z R The distances of the aerodynamic rudder's center of gravity in the x, y, and z directions are considered, taking into account different rudder angles. P This is the distance between the pressure center and the rudder shaft. For propeller torque, z P y P F represents the distance between the propeller's center of gravity and the z-axis. Pi (i = 1, 2) represents the magnitude of the propeller force. For the bow nozzle torque, δ Si (i = 1, 2) represents the bow nozzle deflection angle, F Si (i = 1, 2) represents the magnitude of the bow nozzle force, x S y S z S (i = 1, 2) represents the distances of the bow nozzle in the x, y, and z directions. These are the moments of inertia along the x, y, and z axes, respectively.

[0080] The propulsion constraints of a hovercraft are rudder angle constraints and thrust vector constraints, which are Ω respectively. A,k ∈Θ A,k ,Ω F,k ∈Θ F,k ,in This refers to the range of aerodynamic deflection angles.

[0081] This constrains the propeller thrust and the thrust and deflection angle of the bow nozzle.

[0082] Step 2: Using the mathematical model established in Step 1, design a virtual control law based on an extended state observer;

[0083] Specifically, step 2 includes:

[0084] Step 2-1: Write the angular velocity equation as Form, ψ = [p; q; r], T w =[T p ;T q ;T r This represents the total model uncertainty and total environmental disturbances related to the hovercraft's heel, trim, and yaw. The virtual control law to be designed.

[0085] Step 2-2: Confirm The relationship between the actual control input and the actual control input is denoted as:

[0086]

[0087] Among them, H u = [H1,H2,H3,H4,H5,H6,H7]∈R 3×7 ,

[0088] X = [δ] R ,F P1 ,F P2 ,F S1 sinδ S1 ,F S2 sinδ S2 ,F S1 cosδ S1 ,F S2 cosδ S2 ].

[0089] According to the Lagrange mean value theorem, the aerodynamic rudder torque is transformed as follows:

[0090]

[0091] in, Lagrange expansion point and endpoint δ R At any point between these points, combining the above equation with the angular velocity equation yields...

[0092]

[0093] Combining the angular velocity equation and the propeller and bow nozzle moment equations, the lever arm is extracted to obtain:

[0094]

[0095] Where J = diag(J x J y J z ).

[0096] Steps 2-3: Design a second-order extended state observer using the bandwidth method to compensate for the nonlinear uncertainties of the angular velocity control system and the total environmental disturbances, β 01 ,β 02 For system bandwidth, β p ,β q ,β r The bandwidth control objectives in the p, q, and r directions are respectively to make the angular velocity track the time-varying reference signal ψ*(t)=[p*(t),q*(t),r*(t)]. T :

[0097]

[0098] Where, β 01 =2diag(β) p ,β q ,β r ), To compensate for uncertainties and environmental disturbances in the angular velocity control system, a virtual control law is designed using error proportional feedback control.

[0099]

[0100] k p ,k q ,k r This is the proportional feedback gain for tracking error.

[0101] Step 3: Based on the virtual control input designed in Step 2, design a hierarchical allocation strategy. See the schematic diagram for details. Figure 2 ;

[0102] Specifically, step 3 of the design includes:

[0103] Step 3-1: Set the difference between the virtual control input and the actual controller output as the loss function Γ,k. Simulate the time series:

[0104]

[0105] Step 3-2: Design Hierarchy Allocation Step 1. Since the hovercraft has the highest rudder rotation efficiency, prioritizing the allocation of rudders will yield better results. The loss function for allocating rudders is denoted as Γ. 1,k:

[0106]

[0107] At this point, the propeller and bow nozzle thrust vectors from the previous moment are used, so the virtual control variable is...

[0108] Step 3-3: Design hierarchical allocation step 2. Based on the premise of hierarchical allocation step 1, the propeller and bow nozzle have relatively low rotational efficiency. Optimize the propeller and bow nozzle thrust vectors, denoted as the loss function Γ. 2,k :

[0109]

[0110] This indicates that the optimal solution from step 1 of the hierarchical allocation is being used here.

[0111] Steps 3-4: Design the hierarchical allocation step 3. Under the premise of the hierarchical allocation step 2, minimize the thrust of the propeller and the bow nozzle to reduce the energy consumption of the engine and gas turbine. Design the energy minimization function denoted as E. k :

[0112]

[0113] This indicates the optimal solution obtained by using the hierarchical allocation step 2.

[0114] Step 4: Since the efficiency of each propeller is different at different speeds of the hovercraft, a hierarchical allocation weight coefficient is designed to make the hierarchical allocation strategy more in line with the actual propeller allocation during the actual navigation of the hovercraft.

[0115] Specifically, step 5 includes:

[0116] Step 4-1: The lateral and longitudinal velocities of the fully-cushioned hovercraft at time k are u and u, respectively. k v k Since the rudder's efficiency is highest when the hovercraft is sailing at high speed, the rudder's efficiency will increase with increasing lateral speed. Therefore, the weighting coefficient for step 1 of the hierarchical allocation can be set as follows:

[0117]

[0118] Step 4-2: When the hovercraft is traveling at low speeds or at high lateral speeds, the bow nozzle will play an irreplaceable role. The weighting coefficients for step 2 of the hierarchical allocation can be set as follows:

[0119]

[0120] Since fuel consumption should be kept to a minimum at all times, the weighting coefficient for step 3 of the hierarchical allocation should always be 1.

[0121] Step 5: Solve the hierarchical allocation step loss function using the improved multi-objective Golden Eagle optimization algorithm. See the flowchart for the specific algorithm. Figure 3 The Golden Eagle Optimization Algorithm is based on the circling motion of golden eagles during hunting, where each eagle remembers the best position it has visited. Golden eagles exhibit two tendencies: one is to attack the current prey, and the other is to cruise in search of better prey. The multi-objective Golden Eagle Optimization Algorithm can solve the Pareto optimal solution for multi-objective functions faster and more accurately.

[0122] Specifically, step 5 includes:

[0123] Step 5-1: Confirm the objective function of the multi-objective Golden Eagle algorithm as F = {{Γ} 1,k , Γ 2,k}, E k The weight coefficients of the objective function are {{η}. 1,k η 2,k The cruising vector in the Golden Eagle optimization algorithm is},1}, Attack vector is Prey is the optimal variable

[0124] Initialize the number of golden eagles and estimate the loss function Γ. 1,k and Γ 2,k and the energy minimization function E k The value of the initialization function is used to initialize the Golden Eagle memory bank and the attack coefficient p. a and cruise coefficient p c The attack coefficient increases with the number of iterations, while the cruise coefficient decreases with the number of iterations. This simulates the golden eagle's transition from cruise to attack, moving from its initial position towards its target prey.

[0125] Step 5-2: For each iteration k, update p a and p c The value of p is calculated, and the crowding distance of existing memory bank members is calculated, causing the Golden Eagle to gradually switch from cruise mode to attack mode, and p is updated. a and p c The method is as follows:

[0126]

[0127] T represents the total number of iterations, and k represents the current iteration. and These are the initial coefficients. and The final target value. Set the current position to... The terminal is the optimal variable. The attack model of the Golden Eagle can then be represented as:

[0128]

[0129] Step 5-3: The Golden Eagle cruises to move to a new location, described as follows:

[0130]

[0131] Let i be the step size of the golden eagle i in k iterations.

[0132] Combining a hierarchical allocation strategy, the loss function Γ for the new position is calculated according to a certain priority. 1,k ,Γ 2,k E k If the new position is a Pareto non-dominated solution relative to existing library members, and the external library is not full, then the new solution is added to the library. Otherwise, the sparse distance is calculated, and a roulette wheel selection method with sparse distance weights is used to select the eliminated member, replacing it with the new position. This process is iterated until the algorithm requirements are met.

[0133] Compared with the prior art, the present invention has the following beneficial effects:

[0134] 1) In this invention, the problem of previous thrust distribution problems only involved the control of the roll and pitch angular velocities of the hovercraft, ignoring the problem that the fully-cushioned hovercraft is prone to nose-down or tuck-down. This invention first establishes an angular velocity control model for the fully-cushioned hovercraft and uses three-degree-of-freedom angular velocity control to enable the fully-cushioned hovercraft to sail in a stable attitude.

[0135] 2) A vector thrust hierarchy allocation strategy for hovercraft is proposed, which rationally plans the priority allocation order of each thruster and uses a hierarchy allocation strategy and multiple optimization objectives to improve the utilization efficiency of the thrusters, extend the service life of each thruster, and reduce the overall fuel consumption of the hovercraft.

[0136] 3) An improved multi-objective Golden Eagle optimization algorithm is proposed. The goal is to solve the objective function of the thruster hierarchical allocation strategy. The hierarchical allocation weight coefficient is used to optimize the weight of the multi-objective function, thereby improving the computational efficiency of the multi-objective Golden Eagle optimization algorithm.

[0137] This invention addresses the problem of uneven thrust distribution among the thrusters of a fully-cushioned hovercraft by rationally planning the priority allocation order of each thruster and employing a hierarchical allocation strategy and multiple optimization objectives. It fully considers the execution efficiency of each thruster, improving their efficiency and lifespan, and reducing the overall fuel consumption of the hovercraft. The improved multi-objective Golden Eagle optimization algorithm proposed in this invention can obtain the Pareto optimal solution for the hierarchical allocation strategy loss function within a finite number of steps, and distinguishes the levels of the multi-objective functions, thus improving the computational efficiency of the multi-objective Golden Eagle optimization algorithm.

[0138] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for allocating vector thrust levels for a fully cushioned hovercraft based on an improved Golden Eagle optimization algorithm, characterized in that, Includes the following steps: Step 1: Determine the propulsion system layout of the fully cushioned hovercraft, establish the angular mathematical model of the fully cushioned hovercraft, and load the parameters of each thruster; the parameters of the thrusters include: thruster type and number, thruster thrust direction variable range, thrust magnitude variable range, thrust direction change rate range, and thrust magnitude change rate range. in, For air rudder torque, For propeller torque, For the bow nozzle torque, They are respectively moment of inertia of the shaft These are respectively the roll, trim, and turning angular velocity; Step 2: Design virtual control input based on active disturbance rejection control technology ; in, This refers to the environmental disturbances observed by the observer and the overall uncertainty of the system. For tracking error proportional feedback gain; , The desired angular velocity; Step 3: Design a strategy for allocating the thruster hierarchy of the hovercraft; The difference between the virtual control input and the actual controller output is set as the loss function. k-simulation time series: The propulsion constraints of the hovercraft are rudder angle constraints and thrust vector constraints, respectively. ; This is the actual control output of the hovercraft. For virtual control input; This refers to the range of aerodynamic deflection angles. For propeller thrust constraints and bow nozzle thrust and deflection angle constraints; Hierarchical allocation step 1: Since the hovercraft has the highest rudder rotation efficiency, prioritizing the allocation of rudders will yield better results. The loss function for allocating rudders is denoted as... : Hierarchical Allocation Step 2: Given the low rotational efficiency of the propeller and bow nozzle in Hierarchical Allocation Step 1, optimize the thrust vectors of the propeller and bow nozzle, denoted as the loss function. : This indicates the optimal solution using the hierarchical allocation method in step 1; Hierarchical Allocation Step 3: Based on the premise of Hierarchical Allocation Step 2, minimize the thrust of the propeller and bow nozzle to reduce the energy consumption of the engine and gas turbine. Design the energy minimization function, denoted as... : This indicates the optimal solution using the hierarchical allocation step 2; This refers to the magnitude of the propeller force. The magnitude of the force at the bow nozzle; Step 4: Since the efficiency of each propeller is different at different speeds of the hovercraft, design a hierarchical allocation weight coefficient to make the hierarchical allocation strategy more in line with the actual propeller allocation during the actual navigation of the hovercraft. Step 5: Solve the hierarchical allocation step loss function using the improved multi-objective Golden Eagle optimization algorithm; Step 5.1: Confirm the objective function of the multi-objective Golden Eagle algorithm as follows: The weight coefficients of the objective function are The cruise vector in the Golden Eagle optimization algorithm is The attack vector is The prey is the optimal variable. ; Initialize the number of golden eagles and estimate the loss function. and and energy minimization function The value is initialized in the Golden Eagle Memory Database, and the attack coefficient is initialized. and cruise coefficient Among them, the attack coefficient increases with the number of iterations, while the cruise coefficient decreases with the number of iterations, simulating the golden eagle's transition from cruise to attack, and the process of approaching the target prey from the initial position. Step 5.2: For each iteration ,renew and The value is calculated, and the crowding distance of existing memory bank members is determined, causing the Golden Eagle to gradually transition from cruise mode to attack mode, and the update is performed. and The method is as follows: in, This represents the total number of iterations. For the current iteration, and These are the initial coefficients. and Set the final target value; set the current position as... The terminal variable is the optimal variable. Then the attack model of the Golden Eagle is represented as ; Step 5.3: The Golden Eagle cruises to move to a new location, described as follows: in, For Golden Eagle exist The step size that changes in each iteration; By combining a hierarchical allocation strategy, the loss function for the new position is calculated according to a certain priority. , , If the new position is a Pareto non-dominated solution relative to existing library members and the external library is not full, then the new solution is added to the library; otherwise, the sparse distance is calculated, and a roulette wheel with sparse distance weights is used to select the eliminated member and replace it with the new position. This process is iterated until the algorithm requirements are met.

2. The method for allocating the vector thrust hierarchy of a fully cushioned hovercraft based on the improved Golden Eagle optimization algorithm according to claim 1, characterized in that: The fully-cushioned hovercraft in step 1 has two air propellers, each with two synchronized air rudders, and one bow nozzle on each side of the hull. Full turn; in, , For the forces exerted by a single aerodynamic rudder in the x and y directions, ; , The aerodynamic coefficient is obtained from aerodynamic test data of the vertical air rudder; The speed of the rudder jet brought by the propeller wake; The vertical air rudder area; air density; This represents the distance between the center of gravity of the aerodynamic rudder and the x, y, and z directions. This is the distance between the pressure center and the rudder shaft; in, , The position of the propeller's center of gravity is at shaft and Distance between axes; in, This refers to the deflection angle of the bow nozzle; , , The distances of the bow nozzle in the x, y, and z directions; These are the moments of inertia along the x, y, and z axes, respectively.

3. The method for allocating the vector thrust hierarchy of a fully-cushioned hovercraft based on the improved Golden Eagle optimization algorithm according to claim 2, characterized in that: Step 2 specifically involves: The equation for angular velocity is written as form, , The total model uncertainty and total environmental disturbances for the hovercraft's heel, trim, and yaw. The relationship between the actual control input and the actual control input is denoted as: in, ; ； According to the Lagrange mean value theorem, the aerodynamic rudder torque is transformed as follows: in, Lagrange expansion point and endpoints For any point between these points, combining the above equation with the angular velocity equation, we get: Combining the angular velocity equation and the propeller and bow nozzle moment equations, the lever arm is extracted to obtain: in, ; A second-order extended state observer is designed using the bandwidth method to compensate for the nonlinear uncertainties and total environmental disturbances of the angular velocity control system. For system bandwidth, They are respectively The objective of directional bandwidth control is to enable angular velocity to track a time-varying reference signal. : in, , To compensate for uncertainties and environmental disturbances in the angular velocity control system, a virtual control law is designed using error proportional feedback control. This is the proportional feedback gain for tracking error.

4. The method for allocating vector thrust levels for a fully cushioned hovercraft based on the improved Golden Eagle optimization algorithm according to claim 1, characterized in that: Step 4 specifically involves: The weighting coefficients for step 1 of the hierarchical allocation are set as follows: in, For fully cushioned air-cushioned vehicles The lateral and longitudinal velocities at any given moment; The weighting coefficients for step 2 of the hierarchical allocation are set as follows: The weight coefficient for step 3 of the hierarchical allocation is always 1.

Images