An artificial intelligence adaptive generation method for skateboard chassis

Abstract

The present invention discloses an artificial intelligence adaptive generation method for a skateboard chassis, which relates to the technical field of skateboard chassis. The method is to use artificial intelligence to generate a simulation road spectrum, adopt the envelope of each modular chassis-related component to perform matching and shape selection, and perform anti-interference avoidance on the component space installed on the chassis and the motion range space by moving the finite element unit cells; then use the membrane field method to generate a cell unitized topology scheme for the skateboard chassis; finally, through an iterative cycle, by adjusting the connection relationship between the chassis and the different components connected to it, as well as the diversified virtual in-loop data generated by different uses, different design forms are quickly matched to achieve modular decoupling between components. The skateboard chassis generated by the above method has high versatility, ensures the connection reliability of each system, the high load-bearing capacity of the structure, and has high passive safety performance.

Description

Technical Field

[0001] The present invention relates to the technical field of skateboard chassis, and more specifically to an artificial intelligence adaptive generation method for skateboard chassis. Background Art

[0002] The skateboard chassis is an integrated chassis architecture designed for electric vehicles. It integrates the battery system, wire-controlled steering system, wire-controlled braking system, suspension system, etc. into the chassis. Through the reserved electrical and body interfaces, the upper and lower bodies are separated, so that the body and cabin can be replaced according to needs, forming a universal and shareable electric vehicle chassis platform.

[0003] In recent years, the concept of "skateboard chassis" has been favored by various OEMs and parts manufacturers, and its development has become increasingly mature. However, due to the short development time and the limitations of existing technologies, the development of a universal interface for decoupling the upper and lower bodies of the skateboard chassis is difficult and has a long development cycle. To solve the above problems, the Chinese invention patent application with publication number CN115179871 A discloses a sliding chassis system with separated upper and lower bodies, including an upper body and a lower body, and the upper body is connected to the lower body through a matching standardized interface. Through the sliding chassis system and design method with separated upper and lower bodies, this invention can solve the problem of high cost and long cycle caused by the frequent matching development of upper and lower bodies of vehicle manufacturers, and achieve the effect of reusing the development of the lower body and focusing on the development of the diversity of the upper body.

[0004] However, the skateboard chassis still has the following defects: (1) The high degree of integration leads to difficulties in maintenance, heat dissipation, and reduced collision safety factor; (2) The uncertainty of the vehicle body strength puts higher requirements on the load-bearing capacity, strength and stiffness of the chassis structure, and the active and passive safety of the chassis.

[0005] Therefore, there is an urgent need for an automated production method that can take into account multi-dimensional performance and quality requirements in the early stages of development, achieve the separation of different concerns in the chassis design process, and continuously receive feedback on micro-design plans to find a suitable material layout that can take into account various aspects such as performance / reliability / maintenance requirements and solidify the plan for design refinement. Summary of the Invention

[0006] The present invention provides an artificial intelligence adaptive generation method for a skateboard chassis. Artificial intelligence is used to generate a simulated road spectrum, and a dissipative system is constructed by continuously inputting new information into the chassis generation system, so that the skateboard chassis solution can be self-generated and self-optimized in the system.

[0007] The present invention adopts the following technical solutions:

[0008] An artificial intelligence adaptive generation method for a skateboard chassis, characterized by comprising the following steps:

[0009] Step 1: Obtain the vehicle's virtual in-loop operating parameters through operational simulation data of the real map;

[0010] Step 2: Design a feedback mechanism for the self-generating system;

[0011] Step 3: Simplify the vehicle model;

[0012] Step 4: Self-generate simplified chassis components with interfaces;

[0013] Step 5: Further simplify the self-generated chassis analysis basis and generate a two-dimensional membrane force structure;

[0014] Step 6: Optimize the two-dimensional membrane force structure to generate a self-generated chassis prototype composed of finite element unit cells;

[0015] Step 7: By moving the finite element unit cells, anti-interference avoidance is performed on the space of the components installed on the chassis and the space of the motion range;

[0016] Step 8: Based on the stress environment after the new components are fastened and installed, and the intended use of the skateboard chassis after being fastened to the body, a slime mold genetic algorithm is used to generate an integrated spatial topological material for the three major component systems of "body-chassis-drive", and further optimize the generated skateboard chassis solution;

[0017] Step 9: For the skateboard chassis solution generated after the stress domain optimization of the integrated body, perform anti-interference avoidance on the envelope of the body's interior and exterior components, and further optimize the overall topology of the "body-chassis-drive" ternary system under the new boundary conditions;

[0018] Step 10: Iterate steps 3 to 9 until there is no new input to the system and the newly generated chassis solution differs from the previous generation solution by less than 5%.

[0019] In a preferred embodiment, the specific implementation of the above step 2 is as follows: by establishing a ternary structure of "model-view-controller" to achieve separation of concerns of the adaptive program, through the change of the focus of the program view, different optimization methods are used in stages to generate the chassis form that the program considers most suitable under the current view at the current stage.

[0020] Specifically, the model-view in the above ternary structure includes: (1) a simplified vehicle model, with parameterization of basic performance of components; (2) a coupling model between components, with parameterization of action transfer between components; (3) a self-generated simplified chassis force optimization model, with the view focusing on chassis force; (4) a self-generated simplified chassis model finite element optimization, with the view focusing on the topological efficiency of the chassis material space; (5) a self-generated simplified chassis model component-inter-component interference model, with the view focusing on spatial interference avoidance of the model; (6) a form-finding model of the chassis and body frame, with the view focusing on the fusion of the force system with the addition of new components; (7) an integrated optimization model of the chassis and body force structure, with the fusion of multi-view data; the controller determines whether there is new optimization data in each of the above models and views. If so, it controls the corresponding adjustment node, and each model-view corresponds to one adjustment node; if not, it outputs a generation plan.

[0021] In a preferred embodiment, the specific practices of the above step three include: 3.1 using cell unit blocks to decompose hierarchically the various components that need to be added to the optimization analysis system; 3.2 setting interfaces for coupling with other modules for each divided module, and the interfaces include two elements: the spatial positioning point of the interface and the connection direction of the interface; 3.3 skateboard chassis with different uses define various interfaces, and each component that meets the requirements is coupled to the target optimized chassis according to interface matching.

[0022] In a preferred embodiment, the specific steps of the above-mentioned step four are as follows: 4.1 The reference system performs a rectangular envelope on the three-dimensional chassis model matched in the model library according to the existing vehicle model, and positions the front and rear axles with reference to the relevant suspension installation method and the load-bearing purpose input according to the design requirements, thereby determining the suspension interface of the chassis; 4.2 Simplify the chassis model. After simplification, the system only inherits the length, width and wheelbase information of the matching chassis in the original model library.

[0023] In a preferred embodiment, the specific steps of step 5 are as follows: 5.1 Further simplify the generated simplified original chassis and map it into a two-dimensional elastic membrane structure; 5.2 Determine the internal mesh scale of the elastic membrane structure based on the integer common divisor of the rim size of the matching suspension mounted tire, so that the finite element units at each scale level can cover each other; 5.3 Connect the units in 5.2 diagonally and ensure that the standard mesh is symmetrical along the longitudinal axis using the symmetry strategy; 5.4 Fix the mesh boundary size to ensure that the length direction of the beam remains unchanged; 5.5, align the fixed interface meshes of the front and rear suspensions. Anchoring; 5.6 Apply static load under normal use conditions to the grid as a whole, and find the most reasonable material distribution within the anchoring range; 5.7 Adjust the elastic coefficient of the elastic connecting rod so that the outer contour size of the membrane after deformation can meet the overall boundary definition of the skateboard chassis. After the external force loading and the elastic connecting rods of the grid are balanced, record the various membrane forms generated that meet the conditions as the basic selection model for the subsequent optimization program; 5.8 Perform stress and deformation verification on the membrane forms that meet the requirements, and calculate the deformation stroke of each unit in the direction perpendicular to the membrane plane based on the deformed grid.

[0024] In a preferred embodiment, the specific steps of the above step six are as follows: 6.1. Maintain the tensile stress state of the stress membrane, and at the same time remove the inefficient parts in the membrane according to an appropriate threshold; 6.2 Use small-scale finite element unit cells to envelop the tensile membrane remaining after the holes are dug to form a self-generated chassis prototype.

[0025] In a preferred embodiment, the specific process of step eight is as follows: 8.1 Introducing a stress assessment domain calculation matrix into the system, which is composed of finite element unit calculation cells, each calculation cell moves with the movement of the component; 8.2 Introducing a slime mold genetic operator probe into the system, the slime mold probe randomly moves in the stress space domain with a certain probability vector. By establishing the spatial ecology of the slime mold probe, a relatively stable slime mold survival zone will be formed in the high-scoring areas of the force grid system, and the slime mold in the low-probability force areas will not survive and gradually disappear; 8.3 The in-loop operation parameters generated in step 1 are input to the above system to obtain the transient acceleration of the chassis-body system at each moment, and then mapped to stress parameters in each direction of each spatial domain through the adaptive force grid. These parameters generate the movement direction guidance probability of the genetic operator probe.

[0026] It can be seen from the above description of the present invention that, compared with the prior art, the present invention has the following advantages:

[0027] 1. This invention utilizes artificial intelligence to generate simulated road spectra, employing the envelope of each modular chassis-related component for matching and shape selection. Then, a membrane field method is used to generate a cellular unitized topology for the skateboard chassis. Finally, through an iterative cycle, by adjusting the connection relationships between the chassis and its connected components, as well as the diverse virtual in-the-loop data generated based on different uses, different design forms are rapidly matched, achieving modular decoupling between components. The skateboard chassis generated using this method is highly versatile, ensuring reliable connection between various systems, high structural load-bearing capacity, and superior passive safety performance.

[0028] 2. The present invention uses the "Model-View-Controller (MVC)" architecture pattern to divide the application into three main component groups: model, view and controller, and realizes multi-dimensional optimization of the design scheme through the cyclic activation and iteration of the three components. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 This is the running path diagram of the target skateboard chassis vehicle of the present invention.

[0030] Figure 2 This is a schematic diagram of the external interface positioning and matching of the power component of the present invention.

[0031] Figure 3 This is a program view conversion diagram of the present invention.

[0032] Figure 4 This is a flow chart of the model-view-controller ternary structure of the present invention.

[0033] Figure 5 It is a schematic diagram of the decomposition of the subdivision level of the vehicle model of the present invention.

[0034] Figure 6 This is a schematic diagram of the suspension interface of a further simplified chassis model of the present invention.

[0035] Figure 7 Schematic diagram of the chassis suspension interface redefined by the present invention.

[0036] Figure 8 Schematic diagram of the self-generated chassis prototype of the present invention.

[0037] Figure 9 Schematic diagram of finite element cell avoidance of the present invention.

[0038] Figure 10 Schematic diagram of the stress assessment domain calculation matrix and slime mold probe of the present invention.

[0039] Figure 11 The system of the present invention generates a schematic diagram of material topology avoidance space based on probability iteration. DETAILED DESCRIPTION

[0040] The following describes specific embodiments of the present invention with reference to the accompanying drawings. Numerous details are provided below to provide a comprehensive understanding of the present invention, but those skilled in the art will appreciate that the present invention can be practiced without these details. Well-known components, methods, and processes are not described in detail below.

[0041] This embodiment provides an artificial intelligence adaptive generation method for a skateboard chassis, comprising the following steps:

[0042] Step 1: Obtain the vehicle's virtual in-loop operating parameters through operational simulation data of the real map.

[0043] like Figure 1 Numbers 1-10 are the target skateboard-based transport vehicles, with vehicle number 8 representing a freight vehicle and vehicles numbered 1, 2, and 10 representing public buses. The virtual environment program assigns these vehicles various operational tasks based on test requirements, such as transporting as many passengers as possible from a departure point to a destination, or efficiently transporting express parcels. The program calculates the optimal operating path for each vehicle based on the requirements. After obtaining the operating path, the program records the in-loop parameters of the virtual vehicle during operation, such as the vehicle's speed v along the path (vehicle dynamic performance simulation speed data), heading angle β (generated lateral acceleration data), and the slope α (operational slope change data).

[0044] For skateboard chassis applications with off-road requirements (such as flatbed truck chassis used in special mines), a refined off-road simulation slope system can be obtained through interpolation calculation based on real map elevation data. The Urban Dynamometer Driving Schedule (UDDS) developed by the U.S. Environmental Protection Agency (EPA) is used as a reference for road cycle conditions to correct simulation data to prevent the data from being too idealized, and then off-road vehicle path planning can be carried out to obtain relevant in-loop virtual data.

[0045] According to the in-loop data, the power components in the database are selected and the components that meet the chassis self-generation program are selected as the edge interface module for external input. Figure 2 The details are as follows:

[0046] The maximum power (Pmax) of the drive motor must meet the power requirements at the highest vehicle speed (Pv), the power at the maximum climbing grade (Pp), and the power requirements based on the acceleration time (Pt), that is: Pmax ≥ max[Pv, Pp, Pt].

[0047]

[0048] In the above formula, V max : Maximum speed; η T: transmission efficiency; m: vehicle curb mass (average value specified in relevant standards); f: rolling resistance coefficient (variable depending on the components); C D : air resistance coefficient; A : frontal area (average value of benchmark models); α max : Maximum slope; V t : Vehicle acceleration final speed; V p : vehicle climbing speed; t t : Vehicle acceleration time.

[0049] Select the peak power of the drive motor according to Pmax, and then calculate the rated power. The relationship between the peak power and rated power of the motor is: Ppeak = λPrated, where λ is the motor overload factor.

[0050] The power calculation of the hub motor or multi-motor system is the sum of the motor operating conditions matched to the vehicle. In the road conditions required by uphill and turning sections, since the hub motors with four or more wheel matching transmissions cannot meet the peak power conditions of all motors, it is necessary to select and match them according to the above formula combined with the motor characteristics.

[0051] Step 2: Design the feedback mechanism of the self-generating system.

[0052] By establishing a "model-view-controller" ternary structure, the focus of the adaptive program is separated. By changing the focus of the program view, different optimization methods are used in stages to generate the most suitable chassis form for the current view. The program view conversion is as follows: Figure 3 As shown. The model-view in the above ternary structure refers to Figure 4 , including: (1) vehicle simplified model, corresponding to the basic performance parameterization of components; (2) inter-component coupling model, corresponding to the parameterization of the action transfer between components; (3) self-generated simplified chassis force optimization model, focusing on the chassis force; (4) self-generated simplified chassis model finite element optimization, focusing on the topological efficiency of the chassis material space; (5) self-generated simplified chassis model component interference model, focusing on the spatial interference avoidance of the model; (6) chassis and body frame form-finding model, focusing on the integration of the force system with the addition of new components; (7) chassis body force structure integrated optimization model, integrating multi-view data. The controller determines whether there is new optimization data in the upper model and view. If so, it controls the corresponding adjustment node. Each model-view corresponds to an adjustment node; if not, it outputs the generation solution.

[0053] Step 3: Simplify the vehicle model, refer to Figure 5 .

[0054] 3.1 The hierarchical decomposition using cell blocks requires the addition of various components of the optimized analysis system.

[0055] 3.2 Set interfaces for each divided module to couple with other modules, refer to Figure 2 ,The interface contains two basic elements: (1) the spatial positioning point of the interface; (2) the connection direction of the interface.

[0056] Skateboard chassis with different uses can have different interfaces defined, and all parts that meet the requirements can be coupled to the target optimized chassis according to interface matching. For example:

[0057] Four-wheel drive chassis interface: 4 front, rear, left and right corner drive definition interfaces. These 4 interfaces can be configured with a pair of front and rear corner drive engines of different models. Under different road conditions, the 4 interfaces apply different vector direction forces and different vibrations to the chassis under the torque power configuration strategy. For interface definitions, please refer to Figure 6 .

[0058] The system can establish interface description methods under various operating conditions for strongly coupled mechanical control systems, with the aim of controlling driving energy efficiency under safety constraints and effectively improving the space management level of the entire vehicle by optimizing the chassis topology.

[0059] The system uses a longitudinal / lateral / vertical integrated coordinated control architecture to achieve a multi-objective and multi-constraint integrated coordination mechanism for distributed drive electric vehicles. Figure 10 The force clouds in different XYZ directions in the lower left corner can simulate the complex chassis force conditions of double lane shifting conditions (slight drift) under low adhesion conditions in the system.

[0060] Since the skateboard chassis integrates modules such as the power system / suspension / brake / steering, it is similar to a non-load-bearing chassis. However, the latter lacks the characteristics of the skateboard chassis' hardware and software decoupling development and standard interface, which is very different from traditional chassis such as non-load-bearing chassis. Therefore, the design of the standard interface will greatly affect the subsequent chassis solution generation of the system. Therefore, step two is required to control the adjustment nodes (2 / 3 / 4 / 5 / 6 / 7) to adjust the model view, so that the program optimization direction returns to the power component selection step, that is, the inter-component coupling model (parameterization of the action transfer between components).

[0061] Step 4: Generate simplified chassis components with interfaces.

[0062] The reference system performs rectangular enveloping based on the chassis 3D model matched in the model library of the existing vehicle model, and locates the front and rear axles according to the relevant suspension installation method and the load-bearing purpose input according to the design requirements, thereby determining the chassis suspension interface. Figure 7 . Figure 7In the simulation, the chassis being learned is the front and rear axles of the air suspension, and the suspension force-bearing mounting interface is a spatial range above the wheel arch. To increase the diversity of skateboard chassis generation solutions, it is necessary to further simplify the model and reduce the initial constraints on the force-bearing interface. Therefore, after simplification, the system only inherits the matching chassis length, width, and wheelbase information from the original model library (this information is closely related to the chassis's purpose and does not affect the reasonable degree of freedom of the system's subsequent self-generated solutions).

[0063] Step 5: Further simplify the self-generated chassis analysis basis and generate a two-dimensional membrane force structure.

[0064] 5.1 The generated simplified original chassis is further simplified and mapped into a two-dimensional elastic membrane structure.

[0065] 5.2 To ensure that the finite element units at each scale level can cover each other, the internal grid scale of the elastic membrane structure is determined based on the integer common divisor of the rim size of the matching suspension mounted tire. For example, in this embodiment, the rim radius is 210mm, and 42mm*42mm is taken as the basic unit of the elastic membrane grid. Figure 6 .

[0066] 5.3 In order to establish an effective lateral mechanical feedback system, the system diagonally connects the cells described in 5.2 and ensures that the standard grid is symmetrical along the vertical axis using a symmetry strategy.

[0067] 5.4 Fix the grid boundary size to ensure that the length of the beam remains unchanged.

[0068] 5.5 Anchor the fixed interface grid of the front and rear suspension, see Figure 6 Round sign in .

[0069] 5.6 Since the analysis of all chassis structural components is applicable and based on static stress analysis, static loads under normal use conditions, such as gravity, driving lateral force, and driving transverse force, are applied to the entire grid. As constraints are added, the optimization work will find the most reasonable material distribution within this circled range.

[0070] 5.7 Adjust the elastic coefficient of the elastic connecting rod so that the outer contour size of the membrane after deformation can meet the overall boundary definition of the skateboard chassis. After the external force loading and the elastic connecting rods of the grid reach equilibrium, record the various membrane shapes generated that meet the conditions as the basic selection model for the subsequent optimization program.

[0071] 5.8 The stress and deformation of the membrane morphology that meets the requirements are checked, and the deformation stroke of each unit in the direction perpendicular to the membrane plane is calculated using the deformed grid as the unit. This shows that the spatial tangential stress is large and the material utilization efficiency is low.

[0072] Step 6: Optimize the two-dimensional membrane force structure to generate a self-generated chassis prototype composed of finite element unit cells.

[0073] 6.1 Maintain the tensile stress state of the stress membrane while removing the inefficient parts of the membrane according to the appropriate threshold.

[0074] 6.2 Using small-scale finite element cells ( Figure 8 In this example, the unit size is 21mm*21mm*21mm) and the remaining tensile membrane enveloping the hole forms the prototype of the self-generated chassis.

[0075] 6.3 Due to the deformation of the membrane force matrix in the height direction Z, the generated chassis model will present a 3D spatial layout, which provides a form-finding basis for the preliminary shape of the stamped chassis frame, such as Figure 8 Because the weight of the battery in the force definition is between the front and rear axles, it presents a sunken shape, which is very convenient for guiding the form-finding system to avoid space when arranging battery components in the next step (step seven).

[0076] Step 7: By moving the finite element cells, anti-interference avoidance is performed on the space of components installed on the chassis and the space of the motion range.

[0077] 7.1 Figure 9 The dark-colored trajectory line (except the tire) represents the movement trajectory of the finite element cell during avoidance. The avoidance direction system preferentially selects the main force direction of the membrane unit cell in step five of this part of the finite element unit.

[0078] 7.2 In step five, the connection between membrane cells and finite element cells is recorded using the system's built-in membrane cell-finite element cell relationship data table.

[0079] 7.3 Since the chassis needs to be installed with many components, the avoidance form-finding function in step 7 will be called repeatedly in the iterative loop.

[0080] Step 8: Based on the stress environment after the new components are fastened and installed, and the intended use of the skateboard chassis after being fastened to the body, a slime mold genetic algorithm is used to generate an integrated spatial topological material for the three major component systems of "body-chassis-drive", further optimizing the generated skateboard chassis solution.

[0081] 8.1 Figure 10 First, the stress assessment domain calculation matrix is ​​introduced into the system. It is composed of finite element computational cells, each of which can move with the movement of the component. The different grayscales in the figure represent the stress level in the space at a specific moment. It can be seen that the internal stress is high at the junction of the body and chassis. Due to the vehicle's forward acceleration, there is a certain amount of stress in the front of the body.

[0082] 8.2 Introduce slime mold genetic operator probes into the system, such as Figure 10 The slime mold probe indicated by the middle arrow moves randomly in the stress space domain with a certain probability vector. The operator survives for a long time in places with large stress domains and has a certain proportion of reproduction probability. By establishing the spatial ecology of the slime mold probe, a relatively stable slime mold survival zone will be formed in the high-scoring areas of the stress grid system, and the slime mold cannot survive in the low-probability stress area and gradually disappears.

[0083] 8.3 By inputting the re-loop operation parameters generated in step 1 into the above system, the transient acceleration of the chassis-body system at each moment can be obtained, and then mapped into stress parameters in various directions of each spatial domain through an adaptive force grid. These parameters can generate the movement direction guidance probability of the genetic operator probe.

[0084] Step 9: For the skateboard chassis solution generated after the stress domain adjustment and optimization of the integrated body, perform envelope anti-interference avoidance for the body interior and exterior components, and further optimize the overall topology of the "body-chassis-drive" ternary system under the new boundary conditions.

[0085] like Figure 11 As shown in the figure, the system includes the passenger doors, window glass, interior guardrails, front and rear lights, drive modules, etc. formed according to the arrangement of passenger seats, which will interfere with the force system. The system will avoid these spaces when generating the material topology based on probability iteration.

[0086] At the same time, it can be seen that since the body structure is included in the calculation, a lot of materials in the chassis are removed and the body is used to assist in bearing the force, which can further lightweight the skateboard chassis structure.

[0087] Step 10: Iterate steps 3 to 9 until there is no new input to the system and the newly generated chassis solution differs from the previous generation solution by less than 5% (movement or removal of finite element cells).

[0088] Since the skateboard chassis is designed to carry a variety of vehicle bodies, the loop can quickly match different design forms and achieve modular decoupling between components by adjusting the connection relationship between the chassis and the different components connected to it, as well as the diverse virtual in-the-loop data generated by different uses.

[0089] The above is only a specific implementation of the present invention, but the design concept of the present invention is not limited to this. Any non-substantial changes to the present invention using this concept shall be deemed as an infringement of the protection scope of the present invention.

Claims

1. An artificial intelligence adaptive generation method for a skateboard chassis, characterized in that: The following steps are involved: Step 1: Obtain the vehicle's virtual in-loop operating parameters through operational simulation data of the real map; Step 2: Design a feedback mechanism for the self-generating system; Step 3: Simplify the vehicle model; Step 4: Self-generate simplified chassis components with interfaces; Step 5: Simplify the self-generated chassis analysis basis and generate a two-dimensional membrane force structure. The specific steps of step 5 are as follows: 5.1 Simplify the generated simplified original chassis and map it into a two-dimensional elastic membrane structure; 5.2 Determine the internal mesh scale of the elastic membrane structure based on the integer common divisor of the rim size of the matching suspension-mounted tires, so that the finite element units at each scale level can cover each other; 5.3 Connect the units in 5.2 diagonally, and ensure that the standard mesh is symmetrical along the longitudinal axis using a symmetry strategy; 5.4 Fix the mesh boundary size to ensure that the length of the beam remains unchanged; 5.5 Anchor the fixed interface mesh of the front and rear suspensions; 5.6 Apply static loads under normal use to the entire grid and find the most reasonable material distribution within the anchoring range; 5.7 Adjust the elastic coefficients of the elastic connecting rods so that the outer contour dimensions of the membrane after deformation can meet the overall boundary definition of the skateboard chassis. After the external force loading and the elastic connecting rods of the grid reach equilibrium, record the membrane morphology generated when the conditions are met as the basis for the subsequent optimization process. 5.8 Check the stress and deformation of the membrane shape that meets the requirements, and calculate the deformation stroke of each unit in the direction perpendicular to the membrane plane based on the deformed grid; Step 6: Optimize the two-dimensional membrane stress structure to generate a self-generated chassis prototype composed of finite element unit cells. The specific steps of Step 6 are as follows: 6.

1. Maintain the tension stress state of the stress membrane while removing the inefficient portion of the membrane according to an appropriate threshold; 6.

2. Use small-scale finite element unit cells to envelop the tension membrane remaining after the hole is dug to form the self-generated chassis prototype; Step 7: By moving the finite element unit cells, anti-interference avoidance is performed on the space of the components installed on the chassis and the space of the motion range; Step 8: Based on the stress environment after the new components are fastened and installed, and the intended use of the skateboard chassis after being fastened to the vehicle body, a slime mold genetic algorithm is used to generate an integrated spatial topological material for the "body-chassis-drive" three-component system, and the generated skateboard chassis solution is optimized; Step 9: For the skateboard chassis solution generated after the stress domain optimization of the integrated body, perform anti-interference avoidance on the envelope of the body's interior and exterior components, and optimize the overall topology of the "body-chassis-drive" ternary system under the new boundary conditions. Step 10: Iterate steps 3 to 9 until there is no new input to the system and the newly generated chassis solution differs from the previous generation solution by less than 5%.

2. The method for adaptively generating a skateboard chassis using artificial intelligence according to claim 1, wherein: The specific implementation of step 2 is as follows: by establishing a "model-view-controller" ternary structure to achieve separation of concerns for the adaptive program, and by changing the focus of the program view, using different optimization methods in stages to generate the chassis form that the program considers most suitable under the current view at the current stage.

3. The method for adaptively generating a skateboard chassis using artificial intelligence according to claim 2, wherein: The model-view in the ternary structure includes: (1) a simplified vehicle model, with parameterization of basic performance of components; (2) a coupling model between components, with parameterization of action transfer between components; (3) a self-generated simplified chassis force optimization model, with the view focusing on chassis force; (4) a self-generated simplified chassis model finite element optimization, with the view focusing on the topological efficiency of the chassis material space; (5) a self-generated simplified chassis model component interference model, with the view focusing on spatial interference avoidance of the model; (6) a form-finding model of the chassis and body frame, with the view focusing on the fusion of the force system with the addition of new components; (7) an integrated optimization model of the chassis and body force structure, with the fusion of multi-view data; the controller determines whether there is new optimization data in each of the above models and views, and if so, controls the corresponding adjustment node, with each model-view corresponding to one adjustment node; if not, outputs a generation plan.

4. The method for adaptively generating a skateboard chassis using artificial intelligence according to claim 1, wherein: The specific practices of step three include: 3.1 using cell unit blocks to decompose hierarchically the various components that need to be added to the optimization analysis system; 3.2 setting interfaces for coupling with other modules for each divided module, and the interfaces include two elements: the spatial positioning point of the interface and the connection direction of the interface; 3.3 defining various interfaces for skateboard chassis with different purposes, and each component that meets the requirements is coupled to the target optimized chassis according to the interface matching.

5. The method for adaptively generating a skateboard chassis using artificial intelligence according to claim 1, wherein: The specific steps of step 4 are as follows: 4.1 The reference system creates a rectangular envelope based on the 3D chassis model matched to the existing vehicle model in the model library. It then positions the front and rear axles based on the relevant suspension installation method and the load-bearing purpose entered in the design requirements, thereby determining the chassis suspension interface. 4.2 Simplify the chassis model. After simplification, the system only inherits the length, width and wheelbase information of the matching chassis in the original model library.

6. The artificial intelligence adaptive generation method for a skateboard chassis according to claim 1, characterized in that: The specific process of step eight is as follows: 8.1 A stress assessment domain calculation matrix is ​​introduced into the system, which is composed of finite element unit calculation cells, and each calculation cell moves with the movement of the component; 8.2 A slime mold genetic operator probe is introduced into the system. The slime mold probe moves randomly in the stress space domain with a certain probability vector. By establishing the spatial ecology of the slime mold probe, a relatively stable slime mold survival zone will be formed in the high-scoring areas of the force grid system, and the slime mold in the low-probability force areas will not be able to survive and gradually disappear; 8.3 The system in 8.2 is input with the in-loop operation parameters generated in step 1 to obtain the transient acceleration of the chassis-body system at each moment, and then mapped to stress parameters in each direction of each spatial domain through the adaptive force grid. These parameters generate the movement direction guidance probability of the genetic operator probe.

Images