A layered optimization-based amphibious unmanned platform propulsion system configuration design method

Abstract

The application discloses a layered optimization-based amphibious unmanned platform propulsion system configuration design method and belongs to the technical field of underwater unmanned equipment design and optimization. In view of defects of the prior art, the application obtains required three-axis resultant force and moment of a platform by reverse dynamics calculation based on a preset task trajectory and water dynamic coefficients obtained by CFD numerical simulation, and constructs a theoretical capability requirement boundary; adopts a meta-heuristic algorithm to generate a propulsion system candidate configuration population, takes the requirement boundary as a hard constraint, and quickly eliminates invalid configurations by relying on a generalized force reachable set constructed by a configuration control distribution matrix; for the effective configurations, a QP quadratic programming thrust distribution algorithm and a PID+SMC hybrid closed-loop control strategy are embedded in an inner layer to carry out dynamic trajectory tracking task evaluation, and a comprehensive fitness function integrating total energy consumption and trajectory tracking error is constructed; and based on the fitness value, algorithm iteration optimization is completed, and an optimal propeller configuration is output, which takes into account low energy consumption and high tracking accuracy.

Description

Technical Field

[0001] This invention belongs to the field of underwater unmanned equipment design and optimization technology, specifically relating to a hierarchical optimization-based amphibious unmanned platform propulsion system configuration design method. Background Technology

[0002] Tracked-propeller hybrid amphibious unmanned platforms possess strong mobility and can flexibly switch power output modes according to different operational scenarios in land, water, and transitional zones, exhibiting strong environmental adaptability and maneuverability. Because they can achieve continuous operation across various media and balance the ability to traverse complex terrain with underwater navigation stability, they have broad application prospects in fields such as marine exploration, environmental monitoring, cross-domain inspection, and emergency rescue.

[0003] In this type of amphibious unmanned platform, the thruster is the core power component for underwater motion control. Its spatial layout, installation position, and installation angle directly determine the platform's ability to combine forces and torques during underwater navigation, significantly impacting its maneuverability, attitude stability, and trajectory tracking accuracy. The rationality of the thruster configuration not only affects whether the platform can meet the predetermined motion control requirements but also directly influences the thrust distribution efficiency and energy utilization level of the propulsion system.

[0004] However, most existing amphibious unmanned platform thruster designs adopt an experience-driven, fixed layout, typically only meeting basic navigation and operational needs, lacking systematic optimization designs tailored to specific mission characteristics, trajectory requirements, and power demands. Limited by the number of thrusters, spatial orientation parameters, and layout, platforms often suffer from unreasonable thrust distribution and excessive control redundancy when performing underwater missions, resulting in low overall propulsion system efficiency. In actual operation, an unreasonable thruster configuration forces the platform to consume extra energy to maintain a predetermined trajectory tracking accuracy, leading to energy waste and significantly reducing the platform's underwater endurance and mission execution efficiency. Given fixed thruster performance parameters and mission requirements, optimizing the thruster configuration can effectively improve thrust distribution efficiency and reduce redundant energy consumption in the control process, thereby achieving a synergistic improvement in the energy consumption and trajectory tracking performance of amphibious unmanned platforms during underwater missions. Therefore, optimizing thruster configuration based on actual mission requirements has significant engineering and application value for improving the energy utilization efficiency, trajectory tracking accuracy, underwater endurance, and overall operational performance of amphibious unmanned platforms.

[0005] In the prior art, patent CN113147291A discloses a design scheme for an amphibious cross-medium unmanned vehicle. This unmanned vehicle adopts a water jet propulsion method and can achieve certain cross-medium movement functions. However, it does not carry out multi-objective optimization design for the propulsion system layout in terms of energy consumption and control accuracy. As a result, this type of unmanned vehicle has problems such as high energy consumption and insufficient control accuracy when navigating in water, thus limiting the further improvement of its underwater operation performance.

[0006] Furthermore, patent CN108803634B proposes a method for optimizing thrust allocation in a six-DOF dynamic positioning system for underwater robots based on a genetic algorithm. While this method primarily focuses on dynamic positioning accuracy as the core optimization objective and offers some reference value in thrust allocation control, it fails to incorporate energy consumption into a unified optimization framework. Moreover, it is mainly geared towards fixed-layout underwater robots and is ill-suited for the thruster configuration design requirements of tracked-propeller hybrid amphibious unmanned platforms operating in cross-domain conditions.

[0007] In summary, existing technologies cannot meet the propeller configuration design requirements of tracked-propeller hybrid amphibious unmanned platforms, which need to "combine mission requirements and take into account energy consumption and trajectory tracking accuracy". Therefore, it is urgent to propose a hierarchical propeller configuration optimization method suitable for amphibious unmanned platforms to achieve systematic optimization of propeller configuration parameters and improve the platform's energy utilization efficiency and motion control performance in complex mission scenarios. Summary of the Invention

[0008] To address the problem that existing technologies cannot meet the propeller configuration design requirements of tracked-propeller hybrid amphibious unmanned platforms that "combine mission requirements with consideration of energy consumption and trajectory tracking accuracy," this invention provides a hierarchical optimization-based propulsion system configuration design method for amphibious unmanned platforms. By constructing a theoretical capability requirement boundary and combining it with metaheuristic global search, the method optimizes the propeller orientation parameters, solving the problems of low thrust distribution efficiency, high energy consumption, and poor trajectory tracking accuracy in existing amphibious unmanned platform propeller configurations. Through hard constraint screening, the global optimization space is significantly reduced, effectively avoiding blind searches, and simultaneously considering underwater navigation control accuracy and platform endurance. This fundamentally solves the technical pain points of unreasonable propeller layout, low thrust distribution efficiency, and high navigation energy consumption in traditional amphibious unmanned platforms.

[0009] To achieve the above objectives, the present invention employs the following technical solutions:

[0010] A hierarchical optimization-based configuration design method for amphibious unmanned platform propulsion systems, the method comprising the following steps:

[0011] Step 1: Based on the preset task trajectory and the hydrodynamic coefficients obtained from the numerical simulation of fluid mechanics, the three-axis resultant force and three-axis torque required by the platform are obtained through inverse dynamics calculation, and the theoretical capability requirement boundary is constructed.

[0012] Step 1.1: Based on the preset task trajectory, obtain the platform's position, attitude angle, linear velocity, and angular velocity information in the world coordinate system;

[0013] Step 1.2: Obtain the hydrodynamic coefficients of the platform through computational fluid dynamics numerical simulation, and substitute them into the six-degree-of-freedom spatial dynamic equations to solve the motion parameters in the world coordinate system;

[0014] Step 1.3: Using the attitude transformation Jacobian matrix, the motion parameters in the world coordinate system are mapped and transformed to the body coordinate system to obtain the linear velocity and angular velocity of the platform in the body coordinate system; the mathematical mapping relationship of this attitude transformation is expressed by the following formula:

[0015] in, This represents the platform's pose vector in the world coordinate system. Represents the Jacobian matrix of attitude transformation. This represents the platform's six-degree-of-freedom velocity vector in the body coordinate system;

[0016] Step 1.4: Discretize and differentiate the linear velocity and angular velocity in the body coordinate system using the small step difference method to obtain the linear acceleration and angular acceleration in the body coordinate system;

[0017] Step 1.5: Combining the platform's own physical properties (including mass, moment of inertia, and product of inertia) and hydrodynamic parameters, substitute the linear acceleration and angular acceleration in the body coordinate system into the six-degree-of-freedom spatial dynamics equations to calculate the triaxial resultant force and triaxial torque required by the platform at each moment during trajectory tracking, and determine the theoretical capability requirement boundary. This result characterizes the power and control capability requirements of the platform to complete the preset trajectory, providing a theoretical basis for propulsion system selection and control allocation scheme design. In the solution process, the platform's six-degree-of-freedom spatial dynamics model and the corresponding required generalized force vectors are defined by the following formulas: ; ;

[0018] in, This represents the mass matrix, which includes rigid body inertia and added mass. This represents the six-degree-of-freedom acceleration vector of the platform in the body coordinate system. Represents the Coriolis force and centripetal force matrices. Represents the hydrodynamic damping matrix. This represents the restoring force and restoring moment vector generated by gravity and buoyancy. This represents the generalized force vector required by the platform in the body coordinate system. This represents the generalized force vector of external environmental disturbances. This indicates the resultant force required in the longitudinal direction. This represents the resultant force required in the lateral direction. The required vertical net force , , Represents the coordinate system around the body , , The required torque for the shaft This indicates transpose.

[0019] Step 2: A population of candidate thruster configurations is generated using a metaheuristic algorithm. The theoretical capability requirement boundary is used as a hard constraint. A generalized force reachability set is constructed based on the control allocation matrix corresponding to each candidate thruster configuration. Invalid configurations that do not meet the theoretical capability requirement boundary constraint are eliminated, and valid configurations that meet the theoretical capability requirement boundary constraint are selected.

[0020] The specific operation of step 2 is as follows:

[0021] A heuristic algorithm is used to search for thruster configuration parameters, which are then filtered using theoretical capability requirement boundaries. For a given configuration, combined with closed-loop control, a quadratic programming (QP) thrust allocation problem is solved at each discrete time step of each trajectory to obtain the minimum energy consumption allocation that satisfies the control requirements at that time step. The specific process is as follows:

[0022] Step 2.1: Initialize the particle swarm according to the dimensions of the optimization variables and the preset value range, where each particle is mapped to a candidate thruster configuration of an amphibious unmanned platform, thereby generating a diverse set of initial configuration schemes;

[0023] An evaluation is performed on each configuration in the initial configuration scheme set. To determine whether different configurations meet the theoretical capability requirement boundary under the task conditions, the control allocation matrix corresponding to each configuration is used. Construct the set of generalized forces attainable under constrained thruster conditions; let the thrust output vector of the thruster be... The maximum output amplitude of a single thruster is Based on the control allocation matrix of the candidate configuration And the thrust amplitude constraint of the thruster, to construct the configuration The generalized force can reach the set :

[0024] in, Indicates candidate configuration Thrust output of a certain group of thrusters The generalized force vector that can be generated below; Indicates the first The thrust output of each thruster.

[0025] Extracting the generalized force of demand from the theoretical capability demand boundary By solving a linear programming feasibility problem, it is determined whether there exists a thruster thrust allocation solution that satisfies the above amplitude constraints; when the linear programming problem has a feasible solution, it indicates that... This means that the candidate configuration satisfies the theoretical capability requirement boundary under the current operating conditions; conversely, when the linear programming problem has no feasible solution, it indicates that... This means that the candidate configuration cannot meet the theoretical capability requirement boundary under the current operating conditions.

[0026] Step 2.2: Generalize the demand force for each demand in the theoretical capability demand boundary. The existence of a thruster allocation solution that satisfies the constraints is determined by solving a linear programming feasibility problem.

[0027] Step 2.3: When all demands are generalized If all feasible solutions exist, the candidate configuration is determined to satisfy the theoretical capability requirement boundary constraints and is a valid configuration; otherwise, the configuration is eliminated.

[0028] Step 3: For the effective configuration, embed the quadratic programming (QP) thrust allocation algorithm and the PID+SMC hybrid closed-loop control strategy in the inner layer, carry out dynamic trajectory tracking task evaluation, and construct a comprehensive fitness function that integrates total energy consumption and trajectory tracking error;

[0029] Based on the selection of effective configurations that meet the theoretical capability requirement boundary, dynamic evaluation of the mission trajectory is performed, and the optimal control solution at each sampling moment in the preset mission trajectory is solved. During the solution process, the inner layer constructs a mission evaluation process by integrating a quadratic programming (QP) thrust allocation algorithm with a PID+SMC hybrid closed-loop control strategy. Based on this evaluation process, the actual capability requirement boundary of the platform during actual dynamic tracking is calculated, and the real-time thrust sequence and cumulative thrust consumption index of each thruster along the mission trajectory are obtained through the quadratic programming (QP) thrust allocation algorithm. The formula for the actual capability requirement boundary is as follows:

[0030] in, Represents the proportional control gain matrix. The vector s represents the tracking error s between the platform's current pose or motion state and the preset task trajectory. Represents the integral control gain matrix. Representing the integral variable The tracking error vector at the corresponding time point, Represents the differential control gain matrix. Indicates the rate of change of tracking error. This represents the sliding mode control gain matrix. This represents the boundary layer thickness parameter in sliding mode control. Represents the sliding surface vector. This represents the saturation function, used for boundary layer limiting of the sliding mode control term.

[0031] To comprehensively evaluate the overall energy-saving performance of different thruster configurations in multi-mission scenarios, this invention constructs a hierarchical evaluation mechanism consisting of an inner layer of time-by-time quadratic programming (QP) thrust allocation, trajectory energy consumption accumulation, and an outer layer of multi-trajectory fitness evaluation. Specifically, for any candidate thruster configuration parameters given by the outer layer heuristic algorithm... First, construct the corresponding control allocation matrix. And combine the theoretical capability requirement boundary constraints to determine whether it meets the generalized force requirements corresponding to the preset task trajectory; for feasible configurations that pass the theoretical capability requirement boundary screening, further enter the inner closed-loop trajectory tracking evaluation process.

[0032] In the inner closed-loop trajectory tracking and evaluation process, each preset trajectory is discretized into several time points. At each discrete moment, the PID+SMC closed-loop controller calculates the generalized control force required by the platform based on the current tracking error; given the configuration matrix... Under the condition of minimizing the thruster output energy consumption at that moment, a quadratic programming problem constrained by upper and lower thrust limits is established: ;

[0033] in, Assign thrust vectors to each thruster at this moment. , For the thrust-power fitting parameters of the thruster, To control the allocation of residual weighting matrix; This represents the control allocation matrix corresponding to the current candidate thruster configuration, used to characterize the mapping relationship between the thrust output of each thruster and the platform's six-degree-of-freedom generalized control force.

[0034] The objective function described above can be rearranged into a standard quadratic programming form: ;

[0035] in, , ;

[0036] The timing is obtained using the QP quadratic programming thrust allocation algorithm. Optimal thrust distribution Then, based on the thruster power model, the instantaneous power and discrete energy consumption at that moment are calculated: ; ;

[0037] in, Indicates the first The instantaneous total power of the propulsion system at each discrete sampling moment. Indicates the number of thrusters; Indicates the first Discrete energy consumption corresponding to each discrete sampling time; This represents the time interval between two adjacent discrete sampling moments;

[0038] Therefore, the first The total energy consumption of the task trajectory during the entire execution process is expressed as: ;

[0039] in, This represents the total number of discrete moments for the trajectory. In other words, the energy consumption of a single trajectory is not directly given by a single moment, but is obtained by accumulating the optimal energy consumption results obtained by quadratic programming (QP) thrust allocation for all moments on the trajectory over time.

[0040] Sum the total energy consumption of all trajectories to construct the fitness function of the outer heuristic algorithm: ;

[0041] in, Represents a preset set of trajectories. Represents the weighting coefficients of each trajectory; Indicates the candidate configuration in the 1st The total energy consumption under the trajectory; the outer metaheuristic algorithm updates the individual historical optimal solution and the global optimal solution according to the fitness value until the termination condition is met;

[0042] When all tasks are equally important, one can choose to... To further improve trajectory tracking accuracy, a corresponding tracking error penalty term can be added to the fitness function. The outer heuristic algorithm determines the fitness value based on each candidate configuration. The system updates the historical optimal solution and the global optimal solution of the individual population, and iteratively searches for new thruster configuration parameters until the termination condition is met, finally obtaining the optimal thruster configuration scheme with the minimum overall energy consumption.

[0043] Step 4: Based on the fitness value, complete the algorithm iterative optimization and output the optimal thruster configuration that balances low energy consumption and high tracking accuracy.

[0044] The fitness function also includes a trajectory tracking error penalty term to balance the energy consumption performance of the thruster configuration with the trajectory tracking accuracy.

[0045] The preset task trajectory includes at least one or more of circular trajectory, S-shaped trajectory and L-shaped trajectory, to represent typical underwater motion conditions such as continuous curvilinear motion, path reversal switching and straight-vertical turning.

[0046] The thruster configuration includes the thruster's installation position parameters and installation angle parameters, wherein the installation angle parameters include the azimuth angle and the pitch angle.

[0047] The outer meta-heuristic algorithm is any one of particle swarm optimization, genetic algorithm, or differential evolution algorithm.

[0048] Compared with the prior art, the present invention has the following advantages:

[0049] 1) By using the boundary of the theoretical capability requirement induced by the task as a hard constraint, invalid configurations are quickly eliminated based on the generalized force reachable set, which significantly reduces the search space of the metaheuristic algorithm and effectively avoids the waste of computing resources caused by traditional blind search.

[0050] 2) A hierarchical optimization framework is adopted, coupling an outer-layer metaheuristic global search with an inner-layer quadratic programming (QP) thrust allocation algorithm and PID+SMC hybrid closed-loop control to achieve global optimization of configuration parameters. The outer layer performs global search of configuration parameters for non-convex problems, while the inner layer balances the real-time performance of thrust allocation, trajectory tracking accuracy, and system robustness. This structure, which deeply integrates optimization and control, closely matches actual dynamic operating conditions and truly solves for the optimal platform configuration with the lowest energy consumption.

[0051] 3) Construct a fitness function that integrates total energy consumption and cumulative trajectory tracking error, and conduct multi-dimensional quantitative evaluation of different configuration schemes to effectively avoid the shortcomings of a single optimization objective and ensure that the platform maximizes the overall endurance potential of the system while achieving high-precision operation.

[0052] 4) The optimized 6-propeller xoz planar symmetrical layout effectively solves the problems of low underwater thrust distribution efficiency, high energy consumption and poor tracking accuracy while ensuring amphibious mobility and obstacle crossing ability. It provides theoretical and engineering basis for the design of tracked-propeller composite cross-domain robot propulsion system and can be directly applied to various amphibious unmanned platforms. Attached Figure Description

[0053] Figure 1 This is a flowchart of the hierarchical optimization method for thruster configuration of an amphibious unmanned platform provided in an embodiment of the present invention.

[0054] Figure 2 This is a schematic diagram illustrating the theoretical capability requirement boundary in an embodiment of the present invention;

[0055] Figure 3 This is a schematic diagram of the outer layer metaheuristic algorithm in an embodiment of the present invention;

[0056] Figure 4 This is a schematic diagram of the inner quadratic programming (QP) thrust allocation algorithm in an embodiment of the present invention;

[0057] Figure 5 This is a schematic diagram of the parameterization of the six thrusters configuration for layout optimization in an embodiment of the present invention.

[0058] Figure 6 This is the result of optimization in the embodiments of the present invention. Detailed Implementation

[0059] To gain a deeper understanding of this invention, we will provide a comprehensive and detailed description. However, this invention has various implementations and is not limited to the specific examples listed herein. These examples are presented to enhance a full understanding of the disclosure of this invention.

[0060] The overall flow of the amphibious unmanned platform propulsion system configuration design method based on hierarchical optimization provided by the embodiments of the present invention is as follows: Figure 1 As shown.

[0061] In this embodiment, an amphibious unmanned platform with six symmetrically arranged thrusters is selected as the optimization object. To facilitate configuration design and parameter solving, the thruster installation position parameters and installation angle parameters are used together as optimization variables to establish a six-thruster configuration parameter vector: ;

[0062] in, This indicates the installation position parameters of the thruster in the platform coordinate system. and They represent the first The azimuth and pitch angles of each thruster. Therefore, the six-thrust configuration includes 18 parameters to be optimized, used to characterize the spatial layout of the propulsion system. The parameterized configuration of the six thrusters used for layout optimization in this embodiment is illustrated below. Figure 5 As shown.

[0063] Step 1: Based on the preset task trajectory and combined with the hydrodynamic coefficients obtained from numerical fluid dynamics (CFD) simulation, the three-axis resultant force and three-axis torque required by the platform are obtained through inverse dynamics calculation, and the theoretical capability requirement boundary is constructed.

[0064] Step 1.1: Based on the preset task trajectory, obtain the platform's position, attitude angle, linear velocity, and angular velocity information in the world coordinate system;

[0065] Step 1.2: Obtain the hydrodynamic coefficients of the platform through computational fluid dynamics numerical simulation, and substitute them into the preset dynamic model to solve the motion parameters in the world coordinate system;

[0066] Step 1.3: Using the attitude transformation Jacobian matrix, the motion parameters in the world coordinate system are mapped and transformed to the body coordinate system to obtain the linear velocity and angular velocity of the platform in the body coordinate system; the mathematical mapping relationship of this attitude transformation is expressed by the following formula: ;

[0067] in, This represents the platform's pose vector in the world coordinate system. Represents the Jacobian matrix of attitude transformation. This represents the platform's six-degree-of-freedom velocity vector in the body coordinate system;

[0068] Step 1.4: Discretize and differentiate the linear velocity and angular velocity in the body coordinate system using the small step difference method to obtain the linear acceleration and angular acceleration in the body coordinate system;

[0069] Step 1.5: Combining the platform's own physical properties (including mass, moment of inertia, and product of inertia) and hydrodynamic parameters, substitute the linear acceleration and angular acceleration in the body coordinate system into the six-degree-of-freedom spatial dynamics equations to calculate the triaxial resultant force and triaxial torque required by the platform at each moment during trajectory tracking, and determine the theoretical capability requirement boundary. This result characterizes the power and control capability requirements of the platform to complete the preset trajectory, providing a theoretical basis for propulsion system selection and control allocation scheme design. In the solution process, the platform's multi-rigid-body spatial dynamics model and the corresponding required generalized force vectors are defined by the following formulas: ; ;

[0070] in, This represents the mass matrix, which includes rigid body inertia and added mass. This represents the six-degree-of-freedom acceleration vector of the platform in the body coordinate system. Represents the Coriolis force and centripetal force matrices. Represents the hydrodynamic damping matrix. This represents the restoring force and restoring moment vector generated by gravity and buoyancy. This represents the generalized force vector required by the platform in the body coordinate system. This represents the generalized force vector of external environmental disturbances. This indicates the resultant force required in the longitudinal direction. This represents the resultant force required in the lateral direction. The required vertical net force , , Represents the coordinate system around the body , , The required torque for the shaft This indicates transpose.

[0071] The amphibious unmanned platform is required to execute a standard circular trajectory underwater. Under this condition, the mission trajectory is selected as a circular trajectory, an S-shaped trajectory, and an L-shaped trajectory to represent typical underwater motion conditions such as continuous curvilinear motion, path reversal switching, and straight-line-vertical turning. The hydrodynamic coefficients of the platform under the set cruise trajectory are extracted through CFD simulation. The preset trajectory equations and hydrodynamic coefficients are substituted into the spatial dynamics equations for inverse calculation to obtain the triaxial resultant force and triaxial torque required by the platform at various sampling moments in the body coordinate system while maintaining the preset trajectory and attitude stability. This constructs the theoretical capability requirement boundary for this mission. A schematic diagram of constructing the theoretical capability requirement boundary in this embodiment is shown in Figure 2.

[0072] Step 2: A population of candidate thruster configurations is generated using a metaheuristic algorithm. The theoretical capability requirement boundary is used as a hard constraint. Invalid configurations are quickly eliminated by relying on the generalized force reachability set constructed by the control allocation matrix of each configuration, and the effective configurations that meet the theoretical capability requirement boundary are selected.

[0073] A heuristic algorithm is used to search for thruster configuration parameters, which are then filtered using theoretical capability requirement boundaries. For a given configuration, combined with closed-loop control, a quadratic programming (QP) thrust allocation problem is solved at each discrete time step of each trajectory to obtain the minimum energy consumption allocation that satisfies the control requirements at that time step. The specific process is as follows:

[0074] Step 2.1: Initialize the particle swarm according to the dimensions of the optimization variables and the preset value range, where each particle is mapped to a candidate thruster configuration of an amphibious unmanned platform, thereby generating a diverse set of initial configuration schemes;

[0075] An evaluation is performed on each configuration in the initial configuration scheme set. To determine whether different configurations meet the theoretical capability requirement boundary under the task conditions, the control allocation matrix corresponding to each configuration is used. Construct the set of generalized forces attainable under constrained thruster conditions; let the thrust output vector of the thruster be... The maximum output amplitude of a single thruster is Based on the control allocation matrix of the candidate configuration And the thrust amplitude constraint of the thruster, to construct the configuration The generalized force can reach the set : ;

[0076] Extracting the generalized force of demand from the theoretical capability demand boundary By solving a linear programming feasibility problem, it is determined whether there exists a thruster thrust allocation solution that satisfies the above amplitude constraints; when the linear programming problem has a feasible solution, it indicates that... This means that the candidate configuration satisfies the theoretical capability requirement boundary under the current operating conditions; conversely, when the linear programming problem has no feasible solution, it indicates that... This means that the candidate configuration cannot meet the theoretical capability requirement boundary under the current operating conditions.

[0077] Step 2.2: Generalize the demand force for each demand in the theoretical capability demand boundary. The existence of a thruster allocation solution that satisfies the constraints is determined by solving a linear programming feasibility problem.

[0078] Step 2.3: When all demands are generalized If all feasible solutions exist, the candidate configuration is determined to satisfy the theoretical capability requirement boundary constraints and is a valid configuration; otherwise, the configuration is eliminated.

[0079] Step 3: For the effective configuration, embed the quadratic programming (QP) thrust allocation algorithm and the PID+SMC hybrid closed-loop control strategy in the inner layer, carry out dynamic trajectory tracking task evaluation, and construct a comprehensive fitness function that integrates total energy consumption and trajectory tracking error;

[0080] The specific operation of step 3 is as follows:

[0081] Based on the selection of effective configurations (i.e., particles) that meet the theoretical capability requirement boundary, dynamic evaluation of the mission trajectory is performed, and the optimal control solution at each sampling moment in the preset mission trajectory is solved. During the solution process, the inner layer constructs a mission evaluation process by integrating a quadratic programming (QP) thrust allocation algorithm with a PID+SMC hybrid closed-loop control strategy. Based on this evaluation process, the actual capability requirement boundary of the platform during actual dynamic tracking is calculated, and the real-time thrust sequence and cumulative thrust consumption index of each thruster along the mission trajectory are obtained through the quadratic programming (QP) thrust allocation algorithm. The formula for the actual capability requirement boundary is as follows: ;

[0082] in, Represents the proportional control gain matrix. This represents the tracking error vector relative to the platform's current pose or motion state and the preset task trajectory. Represents the integral control gain matrix. Representing the integral variable The tracking error vector at the corresponding time point, Represents the differential control gain matrix. Indicates the rate of change of tracking error. This represents the sliding mode control gain matrix. This represents the boundary layer thickness parameter in sliding mode control. Represents the sliding surface vector. This represents the saturation function, used for boundary layer limiting of the sliding mode control term.

[0083] To comprehensively evaluate the overall energy-saving performance of different thruster configurations in multi-mission scenarios, this invention constructs a hierarchical evaluation mechanism consisting of an inner layer of time-by-time quadratic programming (QP) thrust allocation, trajectory energy consumption accumulation, and an outer layer of multi-trajectory fitness evaluation. Specifically, for any candidate thruster configuration parameters given by the outer layer heuristic algorithm... First, construct the corresponding thruster configuration matrix. And combine the theoretical capability requirement boundary constraints to determine whether it meets the preset task requirements; for feasible configurations that pass the theoretical capability requirement boundary screening, further enter the inner closed-loop trajectory tracking evaluation process.

[0084] In the inner evaluation, each preset trajectory is discretized into several time points. At each discrete moment, the PID+SMC closed-loop controller calculates the generalized control force required by the platform based on the current tracking error; given the configuration matrix... Under the condition of minimizing the thruster output energy consumption at that moment, a quadratic programming problem constrained by upper and lower thrust limits is established: ; ;

[0085] in, Assign thrust vectors to each thruster at this moment. , For the thrust-power fitting parameters of the thruster, To control the allocation of residual weighting matrix; This represents the control allocation matrix corresponding to the current candidate thruster configuration, used to characterize the mapping relationship between the thrust output of each thruster and the platform's six-degree-of-freedom generalized control force.

[0086] The objective function described above can be rearranged into a standard quadratic programming form: ;

[0087] in, , ;

[0088] At the moment of gain Optimal thrust distribution Then, based on the thruster power model, the instantaneous power and discrete energy consumption at that moment are calculated: ; ;

[0089] in, Indicates the first The instantaneous total power of the propulsion system at each discrete sampling moment. Indicates the number of thrusters; Indicates the first Discrete energy consumption corresponding to each discrete sampling moment; This represents the time interval between two adjacent discrete sampling moments.

[0090] Therefore, the first The total energy consumption of the task trajectory during the entire execution process is expressed as: ;

[0091] in, This represents the total number of discrete moments for the trajectory. In other words, the energy consumption of a single trajectory is not directly given by a single moment, but is obtained by accumulating the optimal energy consumption results obtained by quadratic programming (QP) thrust allocation for all moments on the trajectory over time.

[0092] Sum the total energy consumption of all trajectories to construct the fitness function of the outer heuristic algorithm: ;

[0093] in, Represents a preset set of trajectories. Represents the weighting coefficients of each trajectory; Indicates the candidate configuration in the 1st The total energy consumption under the trajectory; the outer metaheuristic algorithm updates the individual historical optimal solution and the global optimal solution according to the fitness value until the termination condition is met;

[0094] When all tasks are equally important, one can choose to... To further improve trajectory tracking accuracy, a corresponding tracking error penalty term can be added to the fitness function. The outer heuristic algorithm determines the fitness value based on each candidate configuration. The system updates the historical optimal solution and the global optimal solution of the individual population, and iteratively searches for new thruster configuration parameters until the termination condition is met, finally obtaining the optimal thruster configuration scheme with the minimum overall energy consumption.

[0095] Step 4: Based on the fitness value, complete the algorithm iterative optimization and output the optimal thruster configuration that balances low energy consumption and high tracking accuracy.

[0096] When initializing the propulsion system configuration population using a metaheuristic algorithm, the population size is set to 500, the maximum number of iterations is set to 50, and each particle in the population represents a set of spatial installation parameters for 6 thrusters. The thrust output amplitude limit for a single thruster is set as follows. Furthermore, the installation constraints for the thruster's azimuth and pitch angles are... arrive Between these steps, each candidate configuration generated by the algorithm constructs a corresponding generalized force reachability set. If the reachability set of a candidate configuration cannot completely cover the theoretical capability requirement boundary obtained in the first step, the invalid configuration is directly deleted, thereby avoiding subsequent invalid calculations. The outer metaheuristic algorithm in this embodiment is illustrated as follows: Figure 3 As shown.

[0097] For the effective configurations selected through theoretical capability requirement boundary screening, the inner-layer dynamic evaluation is then performed. A PID+SMC hybrid closed-loop control system is built, and the controller parameters are set as follows. and sliding mode control gain In the process of simulating the tracking of a circular trajectory by the control system drive platform, thrust allocation is transformed into a quadratic programming (QP) thrust allocation problem. By strictly limiting and accumulating the total thrust power and trajectory tracking error in the mission time domain, the comprehensive fitness function value of this configuration is calculated. A schematic diagram of the inner-layer quadratic programming (QP) thrust allocation algorithm in this embodiment is shown below. Figure 4 As shown.

[0098] After algorithmic convergence, in this embodiment, the optimal six-thruster configuration obtained by outer-layer optimization is as follows: Figure 6 As shown. The optimization result is: the azimuth angles of the four lateral thrusters are respectively... ; ; ; ;

[0099] The corresponding pitch installation angle is: ;

[0100] To facilitate engineering processing, it will be approximated as [value] during actual manufacturing. The installation angles of the other two thrusters are: ;

[0101] The results show that, under the constraints of this embodiment, when the four main thrusters are installed at a near-horizontal angle and the two auxiliary thrusters are arranged vertically, both planar force synthesis capability and vertical control capability can be taken into account, thereby achieving optimal mission energy consumption performance.

[0102] This embodiment proposes and verifies a hierarchical optimization-based propulsion system configuration design method for amphibious unmanned platforms. By combining a hierarchical architecture of "constructing theoretical capability requirement boundaries to filter ineffective configurations" with "inner-layer dynamic task evaluation and outer-layer global search," the optimal thruster configuration, balancing minimum total energy consumption and maximum trajectory tracking accuracy, was successfully obtained. Through multiple comprehensive evaluations, this optimal configuration significantly reduces overall system energy consumption while maintaining excellent attitude stability and position tracking accuracy under various typical complex task conditions. This invention effectively overcomes the core pain points of traditional fixed layouts, which are limited by experience and have low thrust distribution efficiency. It provides a highly versatile and practical engineering solution for the optimized design of propulsion systems for tracked-propeller hybrid cross-domain robots and various complex amphibious unmanned platforms.

[0103] Contents not described in detail in this specification are prior art known to those skilled in the art. Although illustrative specific embodiments of the invention have been described above to facilitate understanding by those skilled in the art, it should be understood that the invention is not limited to the scope of the specific embodiments. Various modifications are readily apparent to those skilled in the art as long as they fall within the spirit and scope of the invention as defined and determined by the appended claims, and all inventions utilizing the concept of this invention are protected.

Claims

1. A hierarchical optimization-based configuration design method for an amphibious unmanned platform propulsion system, characterized in that, The method includes the following steps: Step 1: Based on the preset task trajectory and the hydrodynamic coefficients obtained from the numerical simulation of fluid mechanics, the three-axis resultant force and three-axis torque required by the platform are obtained through inverse dynamics calculation, and the theoretical capability requirement boundary is constructed. Step 2: A population of candidate thruster configurations is generated using a metaheuristic algorithm. The theoretical capability requirement boundary is used as a hard constraint. A generalized force reachability set is constructed based on the control allocation matrix corresponding to each candidate thruster configuration. Invalid configurations that do not meet the theoretical capability requirement boundary constraint are eliminated, and valid configurations that meet the theoretical capability requirement boundary constraint are selected. Step 3: For the effective configuration, embed a quadratic programming thrust allocation algorithm and a PID+SMC hybrid closed-loop control strategy in the inner layer, conduct dynamic trajectory tracking task evaluation, and construct a comprehensive fitness function that integrates total energy consumption and trajectory tracking error; Step 4: Based on the fitness function, complete the algorithm iterative optimization and output the optimal thruster configuration that balances low energy consumption and high tracking accuracy.

2. The amphibious unmanned platform propulsion system configuration design method based on hierarchical optimization according to claim 1, characterized in that, The specific operation of step 1 is as follows: Step 1.1: Based on the preset task trajectory, obtain the platform's position, attitude angle, linear velocity, and angular velocity information in the world coordinate system; Step 1.2: Obtain the hydrodynamic coefficients of the platform through computational fluid dynamics numerical simulation, and substitute them into the six-degree-of-freedom spatial dynamic equations to solve the motion parameters in the world coordinate system; Step 1.3: Using the attitude transformation Jacobian matrix, the motion parameters in the world coordinate system are mapped and transformed to the body coordinate system to obtain the linear velocity and angular velocity of the platform in the body coordinate system; the mathematical mapping relationship of this attitude transformation is expressed by the following formula: ； in, This represents the platform's pose vector in the world coordinate system. Represents the Jacobian matrix of attitude transformation. This represents the platform's six-degree-of-freedom velocity vector in the body coordinate system; Step 1.4: Discretize and differentiate the linear velocity and angular velocity in the body coordinate system using the small step difference method to obtain the linear acceleration and angular acceleration in the body coordinate system; Step 1.5: Combining the platform's own physical properties and hydrodynamic parameters, substitute the linear acceleration and angular acceleration in the body coordinate system into the six-degree-of-freedom spatial dynamics equations to calculate the triaxial resultant force and triaxial torque required by the platform at each moment during trajectory tracking, and determine the theoretical capability requirement boundary; the formula is: ； ； in, This represents the mass matrix, which includes rigid body inertia and added mass. This represents the six-degree-of-freedom acceleration vector of the platform in the body coordinate system. Represents the Coriolis force and centripetal force matrices. Represents the hydrodynamic damping matrix. This represents the restoring force and restoring moment vector generated by gravity and buoyancy. This represents the generalized force vector required by the platform in the body coordinate system. This represents the generalized force vector of external environmental disturbances. This indicates the resultant force required in the longitudinal direction. This represents the resultant force required in the lateral direction. The required vertical net force , , Represents the coordinate system around the body , , The required torque for the shaft This indicates transpose.

3. The amphibious unmanned platform propulsion system configuration design method based on hierarchical optimization according to claim 2, characterized in that, The specific operation of step 2 is as follows: Step 2.1: Initialize the particle swarm according to the dimensions of the optimization variables and the preset value range, where each particle is mapped to a candidate thruster configuration of an amphibious unmanned platform, thereby generating a diverse set of initial configuration schemes; The candidate thruster configurations in the initial configuration scheme set are evaluated. To determine whether different configurations meet the theoretical capability requirement boundary under mission conditions, the control allocation matrix corresponding to each configuration is used. Construct the set of generalized forces attainable under constrained thruster conditions; let the thrust output vector of the thruster be... The maximum output amplitude of a single thruster is Based on the control allocation matrix of the candidate configuration And the thrust amplitude constraint of the thruster, to construct the configuration The generalized force can reach the set : ； in, Indicates candidate configuration Thrust output of a certain group of thrusters The generalized force vector that can be generated below; Indicates the first The thrust output of each thruster; Extracting the generalized force of demand from the theoretical capability demand boundary By solving a linear programming feasibility problem, it is determined whether a thruster thrust allocation solution that satisfies the above amplitude constraints exists; if a feasible solution exists for the linear programming problem, it indicates that... This means that the candidate configuration satisfies the theoretical capability requirement boundary under the current operating conditions; conversely, when the linear programming problem has no feasible solution, it indicates that... This means that the candidate configuration cannot meet the theoretical capability requirement boundary under the current operating conditions. Step 2.2: Generalize the demand force for each demand in the theoretical capability demand boundary. The existence of a thruster allocation solution that satisfies the constraints is determined by solving a linear programming feasibility problem. Step 2.3: When all demands are generalized If all feasible solutions exist, the candidate configuration is determined to satisfy the theoretical capability requirement boundary constraints and is a valid configuration; otherwise, the configuration is eliminated.

4. The amphibious unmanned platform propulsion system configuration design method based on hierarchical optimization according to claim 3, characterized in that, The specific operation of step 3 is as follows: Based on the selection of effective configurations that meet the theoretical capability requirement boundary, a dynamic trajectory tracking task evaluation is performed, and the optimal control solution at each sampling time in the preset task trajectory is solved. During the solution process, the inner layer constructs a task evaluation process by integrating a quadratic programming thrust allocation algorithm with a PID+SMC hybrid closed-loop control strategy. Based on this evaluation process, the actual capability requirement boundary of the platform during actual dynamic tracking is calculated, and the real-time thrust sequence and cumulative thrust consumption index of each thruster along the task trajectory are obtained through the quadratic programming thrust allocation algorithm. The formula for the actual capability requirement boundary is as follows: ； in, Represents the proportional control gain matrix. This represents the tracking error vector of the platform's current pose or motion state relative to the preset task trajectory. Represents the integral control gain matrix. Representing the integral variable The tracking error vector at the corresponding time point, Represents the differential control gain matrix. Indicates the rate of change of tracking error. This represents the sliding mode control gain matrix. This represents the boundary layer thickness parameter in sliding mode control. Represents the sliding surface vector. This represents the saturation function, used for boundary layer limiting of the sliding mode control term; For any candidate thruster configuration parameters given by the outer heuristic algorithm First, construct the corresponding control allocation matrix. And combine the theoretical capability requirement boundary constraints to determine whether it meets the generalized force requirements corresponding to the preset task trajectory; for feasible configurations that pass the theoretical capability requirement boundary screening, further enter the inner closed-loop trajectory tracking evaluation process. In the inner closed-loop trajectory tracking and evaluation process, each preset trajectory is discretized into several time points. At each discrete moment, the PID+SMC closed-loop controller calculates the generalized control force required by the platform based on the current tracking error; given the configuration matrix... Under the condition of minimizing the thruster output energy consumption at that moment, a quadratic programming problem constrained by upper and lower thrust limits is established, with the following formula: ; ； in, Assign thrust vectors to each thruster at this moment. , For the thrust-power fitting parameters of the propulsion unit, To control the allocation of residual weighting matrix; This represents the control allocation matrix corresponding to the current candidate thruster configuration, used to characterize the mapping relationship between the thrust output of each thruster and the platform's six-degree-of-freedom generalized control force; The objective function described above can be rearranged into a standard quadratic programming form: ； in, , ; The timing is obtained using the QP quadratic programming thrust allocation algorithm. Optimal thrust distribution Then, based on the thruster power model Calculate the instantaneous power and discrete energy consumption at this moment: ； ； in, Indicates the first The instantaneous total power of the propulsion system at each discrete sampling moment. Indicates the number of thrusters; Indicates the first Discrete energy consumption corresponding to each discrete sampling moment; This represents the time interval between two adjacent discrete sampling moments; Therefore, the first The total energy consumption of the task trajectory during the entire execution process is expressed as: ； in, This represents the total number of discrete moments in the trajectory. Sum the total energy consumption of all trajectories to construct the fitness function of the outer heuristic algorithm: ； in, Represents a preset set of trajectories. Represents the weighting coefficients of each trajectory; Indicates the candidate configuration in the 1st The total energy consumption under each trajectory; the outer metaheuristic algorithm updates the individual historical optimal solution and the global optimal solution based on the fitness value until the termination condition is met.

5. The amphibious unmanned platform propulsion system configuration design method based on hierarchical optimization according to claim 4, characterized in that, The fitness function also includes a trajectory tracking error penalty term to balance the energy consumption performance of the thruster configuration with the trajectory tracking accuracy. The preset task trajectory includes at least one or more of circular trajectory, S-shaped trajectory and L-shaped trajectory, to represent typical underwater motion conditions such as continuous curvilinear motion, path reversal switching and straight-vertical turning. The thruster configuration includes the thruster's installation position parameters and installation angle parameters, wherein the installation angle parameters include the azimuth angle and the pitch angle. The outer meta-heuristic algorithm is any one of particle swarm optimization, genetic algorithm, or differential evolution algorithm.

Images