Random feature-based meshless device structure optimization method and equipment

Abstract

The invention provides a meshless device structure optimization method and equipment based on random features, which can be applied to the technical field of computational mathematics and topological optimization. The method comprises the following steps: performing iterative structure optimization on a device by utilizing a displacement field network and a density field network to obtain a device structure optimization result; wherein the step of executing the primary structure optimization comprises the following steps of: adjusting parameters of a previous displacement field network by utilizing previous point density parameters of a plurality of device points of the device to obtain a current displacement field network; inputting respective point coordinates of a plurality of device points of the device into the current displacement field network to obtain respective current point displacement of the plurality of device points; performing parameter adjustment on the previous density field network according to the respective point coordinates of the plurality of device points and the current point displacement to obtain a current density field network; and inputting the respective point coordinates of the plurality of device points into the current density field network to obtain respective current point density parameters of the plurality of device points.

Description

Technical Field

[0001] The present disclosure relates to the fields of computational mathematics and topology optimization technology, and in particular to a gridless device structure optimization method and device based on random features. Background Art

[0002] In the field of engineering topology optimization, traditional device structure optimization is primarily based on mesh discretization methods, using finite element analysis combined with topology optimization algorithms to iteratively adjust material distribution. These methods typically require meshing the design domain, evaluating structural performance by solving mechanical equations, and updating material density or geometric boundaries based on gradient information to ultimately achieve the optimal topology that meets the constraints.

[0003] During the implementation of this disclosure, we discovered that related technologies have at least the following issues: Meshing the device domain requires significant computational resources, but this pre-processing does not contribute to the structural optimization itself, resulting in inefficient utilization of computing resources. Furthermore, related technologies typically optimize device structures using a single physical field, which lacks comprehensiveness and results in poor optimization results. Summary of the Invention

[0004] In view of the above problems, the present disclosure provides a gridless device structure optimization method and apparatus based on random features.

[0005] According to a first aspect of the present disclosure, a gridless device structure optimization method based on random features is provided, comprising: performing iterative structural optimization on a device using a displacement field network and a density field network to obtain a device structure optimization result, wherein the device structure optimization result comprises a target point density parameter and a target point displacement amount of each of a plurality of device points in the device, the target point density parameter indicating a target material filling amount of the device point, and the target point displacement amount indicating a target deformation amount of the device point; wherein performing one structural optimization comprises: adjusting the previous displacement field network using the previous point density parameters of each of the plurality of device points of the device to obtain a current displacement field network; inputting the point coordinates of each of the plurality of device points of the device into the current displacement field network to obtain the current point displacement amount of each of the plurality of device points; adjusting the previous density field network using the point coordinates and the current point displacement amount of each of the plurality of device points to obtain a current density field network; and inputting the point coordinates of each of the plurality of device points into the current density field network to obtain the current point density parameters of each of the plurality of device points.

[0006] According to an embodiment of the present disclosure, the previous point density parameters of each of the multiple device points of the device are used to adjust the parameters of the previous displacement field network to obtain the current displacement field network, including: determining the previous point Young's modulus of each of the multiple device points based on the previous point density parameters of each of the multiple device points and the previous point displacement of each of the multiple device points; determining the previous displacement field network loss value based on the previous point Young's modulus of each of the multiple device points and the previous point displacement of each of the multiple device points; and adjusting the previous displacement field network based on the previous displacement field network loss value to obtain the current displacement field network.

[0007] According to an embodiment of the present disclosure, the displacement field network includes a plurality of cascaded fully connected layers; the parameters of the previous displacement field network are adjusted based on the loss value of the previous displacement field network to obtain the current displacement field network, including: adjusting the parameters of the last fully connected layer in the previous displacement field network based on the loss value of the previous displacement field network to obtain the current displacement field network.

[0008] According to an embodiment of the present disclosure, a previous displacement field network loss value is determined based on the Young's modulus of the previous point of each of the multiple device points and the displacement of the previous point of each of the multiple device points, including: determining an external potential energy loss value based on the Young's modulus of the previous point of each of the multiple device points and the displacement of the previous point of each of the multiple device points; determining an internal potential energy loss value based on the displacement of the previous point of each of the multiple device points; and determining the previous displacement field network loss value based on the external potential energy loss value and the internal potential energy loss value.

[0009] According to an embodiment of the present disclosure, the point coordinates of each of the multiple device points and the current point displacement are used to adjust the parameters of the previous density field network to obtain the current density field network, including: inputting the point coordinates of each of the multiple device points into the previous density field network to obtain the previous point density parameters of each of the multiple device points; determining the previous density field network loss value based on the current point displacement and the previous point density parameters of each of the multiple device points; and adjusting the previous density field network based on the previous density field network loss value to obtain the current density field network.

[0010] According to an embodiment of the present disclosure, a previous density field network loss value is determined based on the current point displacement and the previous point density parameter of each of the multiple device points, including: determining the current device properties of the device based on the current point displacement and the previous point density parameter of each of the multiple device points; determining the current volume based on the previous point density parameter of each of the multiple device points; obtaining the point density reference parameter of each of the multiple device points based on the derivative of the current device properties of the device with respect to the point density parameter and the derivative of the current volume of the device with respect to the point density parameter; and determining the previous density field network loss value based on the current device properties, the current volume and the point density reference parameter.

[0011] According to an embodiment of the present disclosure, the previous density field network loss value is determined based on the current device properties, the current volume and the point density reference parameters, including: determining a first density loss value based on the current device properties and the initial properties of the device; determining a second density loss value based on the current volume and the target volume of the device; determining a third density loss value based on the previous point density parameters of each of the multiple device points and the point density reference parameters of each of the multiple device points; and determining the previous density field network loss value based on the first density loss value, the second density loss value and the third density loss value.

[0012] According to an embodiment of the present disclosure, the density field network includes a plurality of cascaded fully connected layers; the parameters of the previous density field network are adjusted based on the loss value of the previous density field network to obtain the current density field network, including: adjusting the parameters of the last fully connected layer in the previous density field network based on the loss value of the previous density field network to obtain the current density field network.

[0013] According to an embodiment of the present disclosure, the gridless device structure optimization method based on random features also includes: using the current point density parameter as the target point density parameter; obtaining the target Young's modulus of each of the multiple device points based on the target point density parameter and the physical parameters of the target material to be filled in the device; and determining the target point displacement of each of the multiple device points based on the target Young's modulus of each of the multiple device points to obtain the device structure optimization result.

[0014] A second aspect of the present disclosure provides a gridless device structure optimization apparatus based on random features, comprising: a structure optimization module for performing iterative structure optimization on a device using a displacement field network and a density field network to obtain a device structure optimization result, wherein the device structure optimization result includes a target point density parameter and a target point displacement amount of each of a plurality of device points in the device, wherein the target point density parameter indicates a target material filling amount of the device point, and the target point displacement amount indicates a target deformation amount of the device point; wherein the structure optimization module comprises: a first parameter adjustment submodule for adjusting the parameters of a previous displacement field network using the previous point density parameters of each of the plurality of device points of the device to obtain a current displacement field network; a displacement determination submodule for inputting the point coordinates of each of the plurality of device points of the device into the current displacement field network to obtain the current point displacement amount of each of the plurality of device points; a second parameter adjustment submodule for adjusting the parameters of the previous density field network using the point coordinates and current point displacement amount of each of the plurality of device points to obtain a current density field network; and a density determination submodule for inputting the point coordinates of each of the plurality of device points into the current density field network to obtain the current point density parameters of each of the plurality of device points.

[0015] A third aspect of the present disclosure provides an electronic device, comprising: one or more processors; and a memory for storing one or more computer programs, wherein the one or more processors execute the one or more computer programs to implement the steps of the above method.

[0016] The fourth aspect of the present disclosure further provides a computer-readable storage medium having a computer program or instructions stored thereon, which implements the steps of the above method when the computer program or instructions are executed by a processor.

[0017] The fifth aspect of the present disclosure further provides a computer program product, comprising a computer program or instructions, which implement the steps of the above method when executed by a processor.

[0018] According to the disclosed embodiments, an iterative structural optimization of a device is performed using a displacement field network and a density field network. Using device point coordinates as input, the displacement field and density field network parameters are alternately updated, allowing each iteration to simultaneously optimize both material distribution and deformation state. This meshless approach directly processes discrete device points, avoiding the computational overhead of traditional meshing. Through the coordinated optimization of the two networks, the target material loading and target deformation are simultaneously output for each device point, achieving multi-physics coupled optimization of the device structure. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] The above contents and other objects, features and advantages of the present disclosure will become more apparent through the following description of the embodiments of the present disclosure with reference to the accompanying drawings, in which:

[0020] Figure 1 A diagram schematically illustrates an application scenario of a gridless device structure optimization method and apparatus based on random features according to an embodiment of the present disclosure;

[0021] Figure 2 A flowchart of a gridless device structure optimization method based on random features according to an embodiment of the present disclosure is schematically shown;

[0022] Figure 3 A model framework diagram of a gridless device structure optimization method based on random features according to an embodiment of the present disclosure is schematically shown;

[0023] Figure 4 Schematically shows a structural block diagram of a gridless device structure optimization apparatus based on random features according to an embodiment of the present disclosure; and

[0024] Figure 5 A block diagram of an electronic device suitable for implementing a gridless device structure optimization method based on random features according to an embodiment of the present disclosure is schematically shown. DETAILED DESCRIPTION

[0025] Hereinafter, embodiments of the present disclosure will be described with reference to the accompanying drawings. However, it should be understood that these descriptions are merely exemplary and are not intended to limit the scope of the present disclosure. In the detailed description below, for ease of explanation, many specific details are set forth to provide a comprehensive understanding of the embodiments of the present disclosure. However, it is apparent that one or more embodiments may also be implemented without these specific details. In addition, in the following description, descriptions of well-known structures and technologies are omitted to avoid unnecessary confusion of the concepts of the present disclosure.

[0026] The terms used herein are only for describing specific embodiments and are not intended to limit the present disclosure. The terms "comprise," "include," etc. used herein indicate the presence of the features, steps, operations, and / or components, but do not exclude the presence or addition of one or more other features, steps, operations, or components.

[0027] All terms used herein (including technical and scientific terms) have the meanings commonly understood by those skilled in the art unless otherwise defined. It should be noted that the terms used herein should be interpreted as having a meaning consistent with the context of this specification and should not be interpreted in an idealized or overly rigid manner.

[0028] When expressions such as "at least one of A, B, and C, etc." are used, they should generally be interpreted in accordance with the meaning commonly understood by those skilled in the art (for example, "a system having at least one of A, B, and C" should include but is not limited to a system having A alone, B alone, C alone, A and B, A and C, B and C, and / or A, B, C, etc.).

[0029] An embodiment of the present disclosure provides a gridless device structure optimization method based on random features, comprising: utilizing a displacement field network and a density field network to perform iterative structural optimization on a device to obtain a device structure optimization result, wherein the device structure optimization result comprises a target point density parameter and a target point displacement amount of each of a plurality of device points in the device, wherein the target point density parameter indicates a target material filling amount of the device point, and the target point displacement amount indicates a target deformation amount of the device point; wherein performing a structural optimization once comprises: utilizing a previous point density parameter of each of a plurality of device points of the device to adjust the parameters of the previous displacement field network to obtain a current displacement field network; inputting the point coordinates of each of a plurality of device points of the device into the current displacement field network to obtain the current point displacement amount of each of a plurality of device points; adjusting the previous density field network using the point coordinates and the current point displacement amount of each of a plurality of device points to obtain a current density field network; and inputting the point coordinates of each of a plurality of device points into the current density field network to obtain the current point density parameters of each of a plurality of device points.

[0030] Figure 1The application scenario diagram of the random feature-based gridless device structure optimization method and apparatus according to an embodiment of the present disclosure is schematically shown.

[0031] like Figure 1 As shown, the application scenario 100 according to this embodiment may include a first terminal device 101, a second terminal device 102, a third terminal device 103, a network 104, and a server 105. The network 104 is used as a medium for providing a communication link between the first terminal device 101, the second terminal device 102, the third terminal device 103, and the server 105. The network 104 may include various connection types, such as wired or wireless communication links or optical fiber cables.

[0032] A user may use a first terminal device 101, a second terminal device 102, or a third terminal device 103 to interact with a server 105 via a network 104 to receive or send messages, etc. Various communication client applications may be installed on the first terminal device 101, the second terminal device 102, or the third terminal device 103, such as shopping applications, web browser applications, search applications, instant messaging tools, email clients, social platform software, etc. (for example only).

[0033] The first terminal device 101 , the second terminal device 102 , and the third terminal device 103 may be various electronic devices having display screens and supporting web browsing, including but not limited to smart phones, tablet computers, laptop computers, desktop computers, and the like.

[0034] The server 105 may be a server that provides various services, such as a background management server (for example only) that supports websites browsed by users using the first terminal device 101, the second terminal device 102, and the third terminal device 103. The background management server may analyze and process received data such as user requests, and feed back processing results (e.g., web pages, information, or data obtained or generated based on user requests) to the terminal devices.

[0035] It should be noted that the gridless device structure optimization method based on random features provided in the embodiment of the present disclosure can generally be executed by the server 105. Accordingly, the gridless device structure optimization apparatus based on random features provided in the embodiment of the present disclosure can generally be set in the server 105. The gridless device structure optimization method based on random features provided in the embodiment of the present disclosure can also be executed by a server or server cluster that is different from the server 105 and can communicate with the first terminal device 101, the second terminal device 102, the third terminal device 103 and / or the server 105. Accordingly, the gridless device structure optimization apparatus based on random features provided in the embodiment of the present disclosure can also be set in a server or server cluster that is different from the server 105 and can communicate with the first terminal device 101, the second terminal device 102, the third terminal device 103 and / or the server 105.

[0036] It should be understood that Figure 1 The number of terminal devices, networks and servers in the embodiment is merely illustrative. Any number of terminal devices, networks and servers may be provided as required.

[0037] The following will be based on Figure 1 The scene described by Figure 2~Figure 3 A gridless device structure optimization method based on random features in a disclosed embodiment is described in detail.

[0038] Figure 2 The flowchart of the random feature-based gridless device structure optimization method according to an embodiment of the present disclosure is schematically shown.

[0039] like Figure 2 As shown, the gridless device structure optimization method based on random features of this embodiment includes operation S200.

[0040] In operation S200 , an iterative structural optimization is performed on the device using a displacement field network and a density field network to obtain a device structure optimization result.

[0041] According to embodiments of the present disclosure, devices may include physical devices used to study structural mechanics-related properties, such as cantilever arms. These devices can be logically segmented to obtain multiple device points. A displacement field network then represents the displacement of each device point in the device after a force is applied, while a density field network represents the density of a target material at each device point in the device. The target material is the physical material used to fill the device.

[0042] According to an embodiment of the present disclosure, the device structure optimization result includes target point density parameters and target point displacement amounts of each of multiple device points in the device. The target point density parameters indicate the target material filling amount of the device point, and the target point displacement amounts indicate the target deformation amount of the device point.

[0043] According to an embodiment of the present disclosure, operation S200 includes operations S210 to S240.

[0044] In operation S210 , the previous displacement field network is adjusted using the previous point density parameters of each of the plurality of device points of the device to obtain a current displacement field network.

[0045] In operation S220 , the coordinates of each of the plurality of device points of the device are input into a current displacement field network to obtain a current point displacement of each of the plurality of device points.

[0046] In operation S230 , the coordinates of each of the plurality of device points and the displacement of the current point are adjusted to the previous density field network to obtain a current density field network.

[0047] In operation S240 , the point coordinates of each of the plurality of device points are input into the current density field network to obtain current point density parameters of each of the plurality of device points.

[0048] According to an embodiment of the present disclosure, the previous point density parameter is the point density parameter of each device point during the previous structural optimization process. When the current round of structural optimization is the first round, the previous point density parameter is the point density parameter of each device point in the initial state of the device.

[0049] According to an embodiment of the present disclosure, the potential energy currently experienced by the device can be determined based on the previous point density parameters of each of the multiple device points. The above potential energy can cause displacement deviations at the device points. Therefore, the loss value of the previous displacement field network can be determined based on the above potential energy, and the parameters of the previous displacement field network can be adjusted so that the potential energy determined based on the current displacement field network indicates that the device points will not experience displacement deviations.

[0050] According to an embodiment of the present disclosure, the current point displacement represents the current displacement of each device point after an external force is applied to the device under the condition of previous point density parameters of multiple device points.

[0051] According to an embodiment of the present disclosure, the current point density parameter represents the density parameter of each of the plurality of device points on the premise that the displacement of the plurality of device points is ensured to be the displacement of the current point after an external force is applied to the device.

[0052] According to the disclosed embodiments, an iterative structural optimization of a device is performed using a displacement field network and a density field network. Using device point coordinates as input, the displacement field and density field network parameters are alternately updated, allowing each iteration to simultaneously optimize both material distribution and deformation state. This meshless approach directly processes discrete device points, avoiding the computational overhead of traditional meshing. Through the coordinated optimization of the two networks, the target material loading and target deformation are simultaneously output for each device point, achieving multi-physics coupled optimization of the device structure.

[0053] According to an embodiment of the present disclosure, the previous point density parameters of each of the multiple device points of the device are used to adjust the parameters of the previous displacement field network to obtain the current displacement field network, including: determining the previous point Young's modulus of each of the multiple device points based on the previous point density parameters of each of the multiple device points and the previous point displacement of each of the multiple device points; determining the previous displacement field network loss value based on the previous point Young's modulus of each of the multiple device points and the previous point displacement of each of the multiple device points; and adjusting the previous displacement field network based on the previous displacement field network loss value to obtain the current displacement field network.

[0054] According to the embodiments of the present disclosure, Young's modulus is a core parameter that describes the stress-strain relationship of a material during elastic deformation. The strain tensor ϵ can be determined based on the displacement of each of the multiple device points at the previous point. The Young's modulus of the previous point can then be determined based on the density parameter and strain tensor of the previous point at each of the multiple device points.

[0055] According to an embodiment of the present disclosure, based on the Young's modulus of the previous point of each of the multiple device points and the displacement of the previous point of each of the multiple device points, it can specifically include: determining the external potential energy loss value based on the Young's modulus of the previous point of each of the multiple device points and the displacement of the previous point of each of the multiple device points; determining the internal potential energy loss value based on the displacement of the previous point of each of the multiple device points; and determining the previous displacement field network loss value based on the external potential energy loss value and the internal potential energy loss value.

[0056] According to the embodiments of the present disclosure, the external potential energy loss value can be used to represent the loss value of the external potential energy of the device, and the internal potential energy loss value can be used to represent the loss value of the internal potential energy of the device. Based on the Young's modulus of the previous point of each of the multiple device points and the displacement of the previous point of each of the multiple device points, the external potential energy loss value is determined as shown in formula (1):

[0057] (1)

[0058] Where V represents the volume of the device, Indicates the area where the device is located. represents the stress tensor, which can be determined according to Young's modulus E by formula (2):

[0059] (2)

[0060] in, represents Poisson’s ratio, tr(ϵ) represents the trace of the strain tensor ϵ, and I is the identity matrix.

[0061] According to an embodiment of the present disclosure, the strain tensor ϵ can be determined based on the displacement of the previous point, as shown in formula (3):

[0062] (3)

[0063] Where u is the displacement of the previous point.

[0064] According to an embodiment of the present disclosure, the stress tensor is determined After that, the boundary range of the device can be determined according to formulas (4) to (6), and it is necessary to ensure that the device point is within the boundary range:

[0065] (4)

[0066] (5)

[0067] (6)

[0068] Among them, u represents the displacement of the device point, n represents the external normal vector, represents the force on the boundary, represents a known force, is the region where the force boundary conditions are defined, It is the region where the displacement boundary condition is defined and is a subset of the global boundary.

[0069] In solid mechanics, the constraints on device points are mainly achieved through these control equations and boundary conditions: the device points inside the structure must satisfy the stress equilibrium equation ( ), and the device points at the boundary are subject to corresponding boundary conditions according to their area - if they are located at the displacement boundary ( ), then force its displacement to be equal to the given value; if it is located at the force boundary ( ), the surface force (the product of stress and normal vector) must be equal to the known load. These constraints are implemented in finite element or topology optimization by constructing a stiffness matrix and adjusting load vectors to ensure that device point behavior meets physical laws and design requirements.

[0070] According to an embodiment of the present disclosure, the internal potential energy loss value is determined based on the displacement of the previous point of each of the multiple device points, as shown in formula (7):

[0071] (7)

[0072] in, It represents the power density of the surface force on the displacement at that point, that is, the inner product of the external force and displacement per unit area, and dA represents the power density of the surface force on the boundary. The integral of the small area elements on The extent of the region representing the device boundary.

[0073] According to the embodiment of the present disclosure, after determining the external potential energy loss value and the internal potential energy loss value, the previous displacement field network loss value Loss can be determined according to formula (8): u :

[0074] (8)

[0075] According to the embodiments of the present disclosure, by separately calculating the external and internal potential energy losses, the total potential energy of the mechanical system is decomposed and processed, enabling a more accurate reflection of the energy state of the displacement field. The external potential energy correlates with material properties and external force response, while the internal potential energy characterizes the structural deformation energy. The combination of the two enables the loss function to fully describe the mechanical equilibrium conditions, thereby guiding the development of the displacement field network towards physically sound optimization.

[0076] According to the embodiments of the present disclosure, Young's modulus is calculated based on the previous point density parameter and the previous point displacement, establishing an explicit correlation between material distribution and mechanical properties. Using Young's modulus and displacement to calculate network loss values ​​allows parameter adjustments in the displacement field network to reflect changes in the material stiffness distribution, ensuring that displacement field optimization consistently conforms to the mechanical properties of the current material distribution. This physical-quantity-based loss calculation imbues network parameter adjustments with clear physical meaning, improving optimization accuracy.

[0077] According to an embodiment of the present disclosure, the displacement field network includes a plurality of cascaded fully connected layers; the parameters of the previous displacement field network are adjusted based on the loss value of the previous displacement field network to obtain the current displacement field network, including: adjusting the parameters of the last fully connected layer in the previous displacement field network based on the loss value of the previous displacement field network to obtain the current displacement field network.

[0078] According to the embodiments of the present disclosure, the direction for adjusting the parameters of the previous displacement field network can be determined based on the loss value of the previous displacement field network. By fixing all parameters except the last fully connected layer, and adjusting only the parameters of the last fully connected layer according to the adjustment direction, the current displacement field network can be obtained.

[0079] According to the embodiments of the present disclosure, the displacement field network employs a cascaded fully connected layer structure, adjusting only the parameters of the last fully connected layer. This partial parameter adjustment approach preserves the fundamental features extracted by the network's previous layers while efficiently fine-tuning the output results. By constraining the parameter adjustment range, drastic changes in the network structure are avoided, ensuring the stability of the optimization process while reducing computational complexity.

[0080] According to an embodiment of the present disclosure, the point coordinates of each of the multiple device points and the current point displacement are used to adjust the parameters of the previous density field network to obtain the current density field network, including: inputting the point coordinates of each of the multiple device points into the previous density field network to obtain the previous point density parameters of each of the multiple device points; determining the previous density field network loss value based on the current point displacement and the previous point density parameters of each of the multiple device points; and adjusting the previous density field network based on the previous density field network loss value to obtain the current density field network.

[0081] According to the embodiments of the present disclosure, by inputting the coordinates of multiple device points into a previous density field network, it is possible to determine the previous point density parameters for each of the multiple device points in the current device. The previous point density parameters represent the volume percentage of the target material within the spatial range corresponding to each device point in the current state of the device. For example, if the previous point density parameter for a device point is 0.5, it means that 50% of the spatial range corresponding to that device point is filled with the target material.

[0082] According to the embodiments of the present disclosure, based on the current displacement of multiple device points and the previous point density parameters, it is possible to determine the deformation of the device in its current state when subjected to an external force, and based on this deformation, determine the device's previous density field network loss value. The previous density field network loss value can be used to indicate problems and hidden dangers that exist in the device due to the current device density. The larger the previous density field network loss value, the less ideal the current device density is, and the more the device density needs to be adjusted.

[0083] According to an embodiment of the present disclosure, after determining the loss value of the previous density field network, the parameter adjustment direction of the previous density field network can be determined based on the loss value of the previous density field network, and the previous density field network can be adjusted based on the parameter adjustment direction to obtain the current density field network.

[0084] According to the embodiments of this disclosure, by combining the current displacement with the previous density parameter output by the density field to calculate the loss value, a mechanism for the inverse effect of displacement changes on material distribution is established. This feedback regulation enables the density field network to dynamically adjust the material distribution based on changes in the displacement field, forming a bidirectional coupled optimization. By utilizing both coordinate input and displacement feedback, the optimization accuracy of material distribution is improved.

[0085] According to an embodiment of the present disclosure, a previous density field network loss value is determined based on the current point displacement and the previous point density parameter of each of the multiple device points, including: determining the current device properties of the device based on the current point displacement and the previous point density parameter of each of the multiple device points; determining the current volume based on the previous point density parameter of each of the multiple device points; obtaining the point density reference parameter of each of the multiple device points based on the derivative of the current device properties of the device with respect to the point density parameter and the derivative of the current volume of the device with respect to the point density parameter; and determining the previous density field network loss value based on the current device properties, the current volume and the point density reference parameter.

[0086] According to embodiments of the present disclosure, current device properties can be used to represent the deformation of a device when subjected to an external force. Therefore, current device properties can include the device's flexibility. Based on the current displacements of multiple device points, the displacement of each device point when subjected to an external force, i.e., the device's deformation, can be determined. Therefore, the current device properties of the device can be determined based on the current displacements of multiple device points.

[0087] According to an embodiment of the present disclosure, the Young's modulus can also be determined based on the previous point density parameters of the device using an interpolation method, as shown in formula (9):

[0088] (9)

[0089] in, is the current Young's modulus, is the minimum Young's modulus, that is, the Young's modulus of the device when the density of the device point is 0, is the maximum Young's modulus, that is, the Young's modulus of the device when the density of the device point is 1, and p is the penalty coefficient. Preferably, p=3.

[0090] According to an embodiment of the present disclosure, after the Young's modulus is determined based on the previous point density parameters of the device, the current device properties of the device can be represented by the external potential energy loss value of the device.

[0091] According to an embodiment of the present disclosure, the current volume of a device can be determined based on the volumes of a plurality of device points and their respective previous point density parameters, wherein the current volume of the device represents the volume of the target material used to constitute the device. In particular, when the device points are evenly divided, the current volume can be expressed using formula (10):

[0092] (10)

[0093] Among them, ρ i Represents the previous point density parameter of the i-th device point.

[0094] According to the embodiments of the present disclosure, point density reference parameters are obtained by calculating device property derivatives and volume derivatives, decomposing global performance indicators into the impact of local density changes. This sensitivity-based analysis method can identify key material distribution areas and directly link the optimization direction of the density field network to the overall performance goal. The simultaneous consideration of device properties and volume currently achieves a balanced multi-objective optimization.

[0095] According to an embodiment of the present disclosure, the previous density field network loss value is determined based on the current device properties, the current volume and the point density reference parameters, including: determining a first density loss value based on the current device properties and the initial properties of the device; determining a second density loss value based on the current volume and the target volume of the device; determining a third density loss value based on the previous point density parameters of each of the multiple device points and the point density reference parameters of each of the multiple device points; and determining the previous density field network loss value based on the first density loss value, the second density loss value and the third density loss value.

[0096] According to an embodiment of the present disclosure, the first density loss value may be used to represent the loss caused by the current device properties compared to the initial device properties, as shown in formula (11):

[0097] (11)

[0098] Among them, c is the current device attribute, and c0 is the initial attribute.

[0099] According to an embodiment of the present disclosure, the target volume of the device can be determined based on the boundary range of the device before iterative optimization. Based on the current volume and the target volume, the second density loss value can be determined, as shown in formula (12):

[0100] (12)

[0101] According to an embodiment of the present disclosure, based on the previous point density parameters of each of the plurality of device points and the point density reference parameters of each of the plurality of device points, a third density loss value can be determined, as shown in formula (13):

[0102] (13)

[0103] Among them, α is the preset penalty coefficient, is the actual previous point density parameter in the j-th iteration, is the j-th point density reference parameter determined based on the optimization criterion method.

[0104] The process of determining the point density reference parameter is shown in formula (14):

[0105] (14)

[0106] Among them, ρ i+1 Represents the i+1th point density reference parameter, ρ i represents the i-th point density reference parameter, and m and λ are hyperparameters.

[0107] According to an embodiment of the present disclosure, after determining the first density loss value, the second density loss value, and the third density loss value, the previous density field network loss value can be obtained by summing the above three loss values, as shown in formula (15):

[0108] (15)

[0109] Where vf is the target volume of the device, and v is the current volume of the device.

[0110] According to the embodiments of the present disclosure, by decomposing the first, second, and third density loss values, loss terms can be constructed from three aspects: performance deviation, volume constraint, and density distribution, thereby constructing a complete optimization target system. Specifically, the functional performance of the device is guaranteed by the performance deviation loss, the material usage is controlled by the volume constraint loss, and the rationality of the material distribution is ensured by the density distribution loss. Through the weighted combination of the three, multiple aspects such as performance, cost, and manufacturing feasibility are taken into account, avoiding the one-sidedness of single-objective optimization.

[0111] According to an embodiment of the present disclosure, the density field network includes a plurality of cascaded fully connected layers; the parameters of the previous density field network are adjusted based on the loss value of the previous density field network to obtain the current density field network, including: adjusting the parameters of the last fully connected layer in the previous density field network based on the loss value of the previous density field network to obtain the current density field network.

[0112] According to the embodiments of the present disclosure, during the process of parameter adjustment, it is necessary to ensure that the parameters before and after the adjustment meet the constraint conditions. Taking the parameter adjustment of the displacement field network as an example, where the parameters are subject to volume constraints, the constraint conditions can be shown as formula (16):

[0113] (16)

[0114] in, is the state equation, is the volume constraint, f obj (·) is the objective function, i.e., the device properties of the device.

[0115] According to the embodiments of the present disclosure, similar to the displacement field network, the density field network adopts a design that cascades fully connected layers and fine-tunes only the parameters of the final layer. This maintains consistent feature extraction capabilities while efficiently responding to optimization needs. This partial parameter adjustment strategy effectively controls the amplitude of network updates while ensuring optimization freedom, preventing oscillation or non-convergence during the optimization process.

[0116] According to an embodiment of the present disclosure, the gridless device structure optimization method based on random features also includes: using the current point density parameter as the target point density parameter; obtaining the target Young's modulus of each of the multiple device points based on the target point density parameter and the physical parameters of the target material to be filled in the device; and determining the target point displacement of each of the multiple device points based on the target Young's modulus of each of the multiple device points to obtain the device structure optimization result.

[0117] The elastic modulus E is a physical property of the material that describes its stiffness. The greater the stiffness of the material, the harder it is to deform. In structural analysis, E directly affects the stiffness matrix K, thereby indirectly affecting the displacement u, as shown in formulas (17) and (18):

[0118] (17)

[0119] (18)

[0120] Where B is the strain-displacement matrix, which can be determined by the shape function derivative, and D is the constitutive matrix, that is, the material stiffness matrix, which can be determined by E and In addition, according to formula (18), it can be determined that external forces F of different magnitudes will also affect u through the system stiffness matrix K. That is, under the premise that the material type of the target material and the structure of the device remain unchanged, the point displacement of the device point is proportional to the external force applied to the device.

[0121] In particular, if the integral form of continuum mechanics is used, the forces applied to the boundaries is a function of change, then u can be expressed using formula (19):

[0122] (19)

[0123] Among them, A[ ] indicates that is a function of the independent variable. The above functional relationship can be obtained by deducing the equation group composed of formulas (4) to (6).

[0124] According to the embodiments of the present disclosure, the target point density parameter is the adjustment target for adjusting the density parameter of each device point in the device after the current iterative adjustment. That is, after each device point is adjusted to the target point density parameter corresponding to the device point, an iteration of device structural optimization can be completed.

[0125] According to an embodiment of the present disclosure, the volume of the target material present in each device point can be determined based on the target point density parameter, and the new target Young's modulus determined after completing this iterative optimization can be determined based on the physical parameters of the target material to be filled in the device and the volume of the target material.

[0126] According to an embodiment of the present disclosure, based on the target Young's modulus of each of the multiple device points, the elastic deformation generated by the multiple device points in response to external forces is calculated, and the positions of the device points after deformation are compared and subtracted from the positions of the device points before deformation to obtain the target point displacement of each of the multiple device points and obtain the device structure optimization result.

[0127] According to the embodiments of the present disclosure, when the device structure optimization results indicate that after this iteration, the target point displacements of multiple device points are all less than the displacement change threshold compared with those before this iteration, it can be determined that the optimization effect produced by the iterative optimization in the current state is no longer obvious, and continuing the iterative optimization will waste computing resources and reduce the optimization efficiency. Therefore, it can be determined to stop the optimization and use the current point density parameters of the multiple device points as the optimization results in progress.

[0128] According to another embodiment of the present disclosure, after device structure optimization, if the change in the current point density parameters of multiple device points compared to before the current iteration is less than a density change threshold, it can be determined that the adjustments made during the iterative optimization are no longer significant enough. Continuing iterative optimization will waste computing resources and reduce optimization efficiency. Therefore, it is determined that the optimization can be stopped, and the current point density parameters of the multiple device points can be used as the ongoing optimization results. Preferably, the density change threshold can be 0.01.

[0129] According to an embodiment of the present disclosure, when the number of iterations reaches an iteration threshold, the optimization can be stopped and the current point density parameters of the multiple device points can be used as the optimization results. Preferably, the iteration threshold can be 500.

[0130] According to the embodiments of the present disclosure, a complete mapping from design parameters to material properties is achieved by directly using the final optimized density parameter as the target value and inferring the Young's modulus based on the material's physical parameters. The target displacement is then determined through mechanical calculations, forming a verification link from material distribution to structural response, ensuring that the final optimization result meets the dual requirements of material layout and mechanical properties.

[0131] Figure 3 A model framework diagram of a gridless device structure optimization method based on random features according to an embodiment of the present disclosure is schematically shown.

[0132] like Figure 3 As shown, the model consists of two parts: a displacement field network and a density field network. Each is a fully connected neural network composed of several layers. Based on the device's boundary conditions, the device's location can be determined so that device points can be collected in that area. In one embodiment, 15,000 device points can be randomly collected in the device's area.

[0133] The coordinates of multiple device points are input into a displacement field network. The network then processes each of these points to determine the displacement of each of the previous points after the device is subjected to an external force. Based on the previous point density parameters of each of these points, the external and internal potential energies of each device point after the device is subjected to the external force are determined. Based on the magnitude relationship and absolute value difference between the external and internal potential energies, the offset of each of the device points can be determined.

[0134] The above-mentioned offset is used as the loss value of the previous displacement field network, and the parameters of the previous displacement field network are adjusted to obtain the current displacement field network. After obtaining the current displacement field network, the coordinates of multiple device points are again input into the current displacement field network to obtain the current point displacement of each of the multiple device points. When compared with the previous point displacement calculated using the previous displacement field network, the current point displacement calculated using the current displacement field network has a smaller difference between the external potential energy and the internal potential energy. In other words, the offset of each device point is smaller, and the current point displacement obtained is closer to the actual displacement.

[0135] The coordinates of multiple device points are input into a density field network. The density field network processes these points individually to obtain their previous point density parameters. Based on the initial device properties, the optimization criterion method (OC) is used to determine the optimal point density for each device point. The root mean square error (RMS) can be constructed based on the optimized direction and the previous point density parameters. The current point displacement and previous point density parameters of each device point can be used to determine the current device properties and volume.

[0136] Based on the initial and device properties of the device, the target and current volumes of the device, and the root mean square error, the previous density field network loss value can be determined, and the parameters of the previous density field network can be adjusted to obtain the current density field network. After obtaining the current density field network, the coordinates of multiple device points are again input into the current density field network to obtain the current point density parameters of each of the multiple device points.

[0137] The optimization process for the displacement field network and density field network is repeated until the number of iterations reaches the iteration threshold, or the difference between the current point density parameter and the previous point density parameter is less than the density change threshold. The iteration is then terminated. The device structure optimization results are then determined based on the current point density parameters of each of the multiple device points.

[0138] According to the device structure optimization results, multiple device points are filled with target materials according to their respective current point density parameters, so as to obtain devices that meet the device property requirements.

[0139] Based on the above-mentioned random feature-based gridless device structure optimization method, the present disclosure also provides a random feature-based gridless device structure optimization device. Figure 4 The device is described in detail.

[0140] Figure 4 The structural block diagram of the random feature-based gridless device structure optimization apparatus according to an embodiment of the present disclosure is schematically shown.

[0141] like Figure 4 As shown, the random feature-based gridless device structure optimization apparatus of this embodiment includes a structure optimization module 400 .

[0142] Structural optimization module 400 is configured to perform iterative structural optimization on the device using a displacement field network and a density field network, obtaining device structural optimization results. The device structural optimization results include target point density parameters and target point displacements for each of multiple device points within the device. The target point density parameters indicate the target material filling amount at the device point, and the target point displacements indicate the target deformation amount at the device point. In one embodiment, structural optimization module 400 can be configured to perform operation S200 described above, and will not be further described here.

[0143] The structure optimization module 400 includes a first parameter adjustment submodule 410 , a displacement determination submodule 420 , a second parameter adjustment submodule 430 and a density determination submodule 440 .

[0144] The first parameter adjustment submodule 410 is used to adjust the parameters of the previous displacement field network using the previous point density parameters of each of the multiple device points of the device to obtain the current displacement field network.

[0145] The displacement determination submodule 420 is configured to input the coordinates of each of the plurality of device points of the device into the current displacement field network to obtain the current point displacement of each of the plurality of device points.

[0146] The second parameter adjustment submodule 430 is used to adjust the coordinates of each of the multiple device points and the current point displacement to the previous density field network to obtain the current density field network.

[0147] The density determination submodule 440 is configured to input the point coordinates of each of the plurality of device points into the current density field network to obtain the current point density parameters of each of the plurality of device points.

[0148] According to an embodiment of the present disclosure, the structure optimization module 410 includes a previous modulus determination submodule, a displacement loss value determination submodule, and a first parameter adjustment submodule.

[0149] The previous modulus determination submodule is used to determine the previous point Young's modulus of each of the multiple device points based on the previous point density parameters of each of the multiple device points and the previous point displacement of each of the multiple device points.

[0150] The displacement loss value determination submodule is used to determine the previous displacement field network loss value based on the previous point Young's modulus of each of the multiple device points and the previous point displacement of each of the multiple device points.

[0151] The first parameter adjustment submodule is used to adjust the parameters of the previous displacement field network based on the loss value of the previous displacement field network to obtain the current displacement field network.

[0152] According to an embodiment of the present disclosure, the first parameter adjustment submodule includes a first parameter adjustment unit.

[0153] The first parameter adjustment unit is used to adjust the parameters of the last fully connected layer in the previous displacement field network based on the loss value of the previous displacement field network to obtain the current displacement field network.

[0154] According to an embodiment of the present disclosure, the displacement loss value determining submodule includes a first potential energy loss determining unit, a second potential energy loss determining unit, and a displacement loss value determining unit.

[0155] The first potential energy loss determining unit is configured to determine an external potential energy loss value based on the Young's modulus of the previous point of each of the plurality of device points and the displacement of the previous point of each of the plurality of device points.

[0156] The second potential energy loss determination unit is used to determine the internal potential energy loss value based on the previous point displacement of each of the multiple device points.

[0157] The displacement loss value determining unit is used to determine the previous displacement field network loss value based on the external potential energy loss value and the internal potential energy loss value.

[0158] According to an embodiment of the present disclosure, the structure optimization module 410 includes a previous density determination submodule, a density loss determination submodule, and a second parameter adjustment submodule.

[0159] The previous density determination submodule is used to input the point coordinates of each of the multiple device points into the previous density field network to obtain the previous point density parameters of each of the multiple device points.

[0160] The density loss determination submodule is used to determine the previous density field network loss value based on the current point displacement and the previous point density parameter of each of the multiple device points.

[0161] The second parameter adjustment submodule is used to adjust the parameters of the previous density field network based on the loss value of the previous density field network to obtain the current density field network.

[0162] According to an embodiment of the present disclosure, the density loss determination submodule includes a current attribute determination unit, a current volume determination unit, a density reference parameter determination unit, and a density loss determination unit.

[0163] The current attribute determination unit is used to determine the current device attribute of the device based on the current point displacement and the previous point density parameter of each of the multiple device points.

[0164] The current volume determining unit is configured to determine the current volume based on the previous point density parameters of each of the plurality of device points.

[0165] The density reference parameter determination unit is used to obtain point density reference parameters of multiple device points based on the derivative of the current device property of the device with respect to the point density parameter and the derivative of the current volume of the device with respect to the point density parameter.

[0166] The density loss determination unit is used to determine the previous density field network loss value based on the current device properties, the current volume and the point density reference parameters.

[0167] According to an embodiment of the present disclosure, the density loss determining unit includes a first density loss determining subunit, a second density loss determining subunit, a third density loss determining subunit, and a density loss determining subunit.

[0168] The first density loss determining subunit is configured to determine a first density loss value based on current device properties and initial device properties.

[0169] The second density loss determining subunit is configured to determine a second density loss value based on the current volume and the target volume of the device.

[0170] The third density loss determining subunit is configured to determine a third density loss value based on the previous point density parameters of each of the plurality of device points and the point density reference parameters of each of the plurality of device points.

[0171] The density loss determining subunit is configured to determine a previous density field network loss value based on the first density loss value, the second density loss value, and the third density loss value.

[0172] According to an embodiment of the present disclosure, the second parameter adjustment submodule includes a second parameter adjustment unit.

[0173] The second parameter adjustment unit is used to adjust the parameters of the last fully connected layer in the previous density field network based on the loss value of the previous density field network to obtain the current density field network.

[0174] According to an embodiment of the present disclosure, the gridless device structure optimization apparatus based on random features further includes a target parameter determination module, a target modulus determination module, and an optimization result determination module.

[0175] The target parameter determination module is used to use the current point density parameter as the target point density parameter.

[0176] The target modulus determination module is used to obtain the target Young's modulus of each of the multiple device points based on the target point density parameter and the physical parameters of the target material to be filled in the device.

[0177] The optimization result determination module is used to determine the target point displacement of each of the multiple device points based on the target Young's modulus of each of the multiple device points, and obtain the device structure optimization result.

[0178] According to embodiments of the present disclosure, any multiple modules among the structure optimization module 400, the first parameter adjustment submodule 410, the displacement determination submodule 420, the second parameter adjustment submodule 430, and the density determination submodule 440 may be combined into a single module, or any one of these modules may be split into multiple modules. Alternatively, at least part of the functionality of one or more of these modules may be combined with at least part of the functionality of other modules and implemented in a single module. According to embodiments of the present disclosure, at least one of the structure optimization module 400, the first parameter adjustment submodule 410, the displacement determination submodule 420, the second parameter adjustment submodule 430, and the density determination submodule 440 may be at least partially implemented as a hardware circuit, such as a field programmable gate array (FPGA), a programmable logic array (PLA), a system on a chip, a system on a substrate, a system on a package, an application-specific integrated circuit (ASIC), or may be implemented in hardware or firmware through any other reasonable means of circuit integration or packaging, or may be implemented in any one of the three implementation methods of software, hardware, and firmware, or any appropriate combination of any of these. Alternatively, at least one of the structure optimization module 400, the first parameter adjustment submodule 410, the displacement determination submodule 420, the second parameter adjustment submodule 430 and the density determination submodule 440 can be at least partially implemented as a computer program module, and when the computer program module is run, the corresponding function can be performed.

[0179] Figure 5 A block diagram of an electronic device suitable for implementing a gridless device structure optimization method based on random features according to an embodiment of the present disclosure is schematically shown.

[0180] like Figure 5 As shown, the electronic device 500 according to an embodiment of the present disclosure includes a processor 501, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 502 or a program loaded from a storage unit 508 into a random access memory (RAM) 503. The processor 501 may include, for example, a general-purpose microprocessor (e.g., a CPU), an instruction set processor and / or a related chipset and / or a special-purpose microprocessor (e.g., an application-specific integrated circuit (ASIC)), etc. The processor 501 may also include onboard memory for caching purposes. The processor 501 may include a single processing unit or multiple processing units for performing different actions of the method flow according to the embodiment of the present disclosure.

[0181] Various programs and data required for the operation of the electronic device 500 are stored in the RAM 503. The processor 501, ROM 502, and RAM 503 are connected to each other via a bus 504. The processor 501 executes the various operations of the method flow according to the embodiment of the present disclosure by executing the programs in the ROM 502 and / or RAM 503. It should be noted that the programs may also be stored in one or more memories other than the ROM 502 and RAM 503. The processor 501 may also execute the various operations of the method flow according to the embodiment of the present disclosure by executing the programs stored in the one or more memories.

[0182] According to an embodiment of the present disclosure, electronic device 500 may further include an input / output (I / O) interface 505, which is also connected to bus 504. Electronic device 500 may also include one or more of the following components connected to I / O interface 505: an input section 506 including a keyboard, mouse, etc.; an output section 507 including devices such as a cathode ray tube (CRT), liquid crystal display (LCD), and speakers; a storage section 508 including a hard disk; and a communication section 509 including a network interface card such as a LAN card or modem. Communication section 509 performs communication processing via a network such as the Internet. A drive 510 is also connected to I / O interface 505 as needed. Removable media 511, such as a magnetic disk, optical disk, magneto-optical disk, semiconductor memory, etc., is installed in drive 510 as needed, so that computer programs read from the removable media can be installed into storage section 508 as needed.

[0183] The present disclosure also provides a computer-readable storage medium, which may be included in the device / apparatus / system described in the above embodiments, or may exist independently and not be incorporated into the device / apparatus / system. The computer-readable storage medium carries one or more programs, and when executed, implements the method according to the embodiments of the present disclosure.

[0184] According to an embodiment of the present disclosure, a computer-readable storage medium may be a non-volatile computer-readable storage medium, and may include, for example, but not limited to: a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In the present disclosure, a computer-readable storage medium may be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. For example, according to an embodiment of the present disclosure, a computer-readable storage medium may include the ROM 502 and / or RAM 503 described above, and / or one or more memories other than ROM 502 and RAM 503.

[0185] Embodiments of the present disclosure also include a computer program product comprising a computer program containing program code for executing the method shown in the flowchart. When the computer program product is executed in a computer system, the program code is used to cause the computer system to implement the random feature-based gridless device structure optimization method provided in the embodiments of the present disclosure.

[0186] The computer program executes the above functions defined in the system / device of the embodiment of the present disclosure when the computer program is executed by the processor 501. According to the embodiment of the present disclosure, the system, device, module, unit, etc. described above can be implemented by a computer program module.

[0187] In one embodiment, the computer program may be stored on a tangible storage medium such as an optical storage device or a magnetic storage device. In another embodiment, the computer program may be transmitted and distributed in the form of a signal on a network medium, downloaded and installed via the communication portion 509, and / or installed from a removable medium 511. The program code contained in the computer program may be transmitted using any appropriate network medium, including but not limited to wireless, wired, or any suitable combination thereof.

[0188] In such an embodiment, the computer program can be downloaded and installed from a network via the communication section 509, and / or installed from a removable medium 511. When the computer program is executed by the processor 501, the above-described functions defined in the system of the embodiment of the present disclosure are performed. According to the embodiment of the present disclosure, the systems, devices, means, modules, units, etc. described above can be implemented by computer program modules.

[0189] According to an embodiment of the present disclosure, the program code for executing the computer program provided by the embodiment of the present disclosure can be written in any combination of one or more programming languages. Specifically, these computer programs can be implemented using high-level procedural and / or object-oriented programming languages, and / or assembly / machine languages. Programming languages ​​include, but are not limited to, languages ​​such as Java, C++, Python, "C" or similar programming languages. The program code can be executed entirely on the user computing device, partially on the user device, partially on a remote computing device, or entirely on a remote computing device or server. In cases involving a remote computing device, the remote computing device can be connected to the user computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computing device (for example, using an Internet service provider to connect via the Internet).

[0190] The flowcharts and block diagrams in the accompanying drawings illustrate the possible implementation architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present disclosure. In this regard, each box in the flowchart or block diagram can represent a module, program segment, or a part of code, and the above-mentioned module, program segment, or a part of code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in an order different from that marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram or flowchart, and the combination of boxes in the block diagram or flowchart, can be implemented with a dedicated hardware-based system that performs the specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0191] Those skilled in the art will appreciate that the features described in the various embodiments of the present disclosure may be combined and / or coupled in various ways, even if such combinations or couplings are not explicitly described in the present disclosure. In particular, the features described in the various embodiments of the present disclosure may be combined and / or coupled in various ways without departing from the spirit and teachings of the present disclosure. All such combinations and / or couplings fall within the scope of the present disclosure.

[0192] The above describes the embodiments of the present disclosure. However, these embodiments are for illustrative purposes only and are not intended to limit the scope of the present disclosure. Although each embodiment has been described separately above, this does not mean that the measures in each embodiment cannot be advantageously used in combination. Without departing from the scope of the present disclosure, those skilled in the art may make various substitutions and modifications, which should all fall within the scope of the present disclosure.

Claims

1. A gridless device structure optimization method based on random features, characterized in that: The method comprises: Performing iterative structural optimization on the device using the displacement field network and the density field network to obtain a device structural optimization result, wherein the device structural optimization result includes a target point density parameter and a target point displacement of each of a plurality of device points in the device, the target point density parameter indicating a target material filling amount of the device point, and the target point displacement indicating a target deformation amount of the device point; Among them, performing a structural optimization includes: Adjusting the parameters of the previous displacement field network using the previous point density parameters of each of the plurality of device points of the device to obtain a current displacement field network; Inputting the coordinates of each of the plurality of device points of the device into the current displacement field network to obtain the current point displacement of each of the plurality of device points; Adjust the coordinates of each of the plurality of device points and the displacement of the current point to the previous density field network to obtain a current density field network; and The point coordinates of each of the plurality of device points are input into the current density field network to obtain the current point density parameters of each of the plurality of device points.

2. The method according to claim 1, characterized in that The method of adjusting the parameters of the previous displacement field network by using the previous point density parameters of the plurality of device points of the device to obtain the current displacement field network includes: Determining a previous point Young's modulus of each of the plurality of device points based on a previous point density parameter of each of the plurality of device points and a previous point displacement of each of the plurality of device points; Determining a previous displacement field network loss value based on the Young's modulus of each of the previous points and the displacement of each of the previous points; and The parameters of the previous displacement field network are adjusted based on the loss value of the previous displacement field network to obtain the current displacement field network.

3. The method according to claim 2, characterized in that The displacement field network includes a plurality of cascaded fully connected layers; The adjusting parameters of the previous displacement field network based on the loss value of the previous displacement field network to obtain the current displacement field network includes: The parameters of the last fully connected layer in the previous displacement field network are adjusted based on the loss value of the previous displacement field network to obtain the current displacement field network.

4. The method according to claim 2 or 3, characterized in that The determining of the previous displacement field network loss value based on the previous point Young's modulus of each of the plurality of device points and the previous point displacement of each of the plurality of device points comprises: determining an external potential energy loss value based on the Young's modulus of the previous point of each of the plurality of device points and the displacement of the previous point of each of the plurality of device points; determining an internal potential energy loss value based on the displacement of each of the plurality of device points at the previous point; and The previous displacement field network loss value is determined based on the external potential energy loss value and the internal potential energy loss value.

5. The method according to claim 1, wherein The step of adjusting the coordinates of each of the plurality of device points and the displacement of the current point to the previous density field network to obtain the current density field network includes: Inputting the point coordinates of each of the plurality of device points into the previous density field network to obtain the previous point density parameters of each of the plurality of device points; Determining a previous density field network loss value based on the current point displacement of each of the plurality of device points and the previous point density parameter; and The previous density field network is adjusted based on the loss value of the previous density field network to obtain the current density field network.

6. The method according to claim 5, characterized in that The determining of the previous density field network loss value based on the current point displacement of each of the plurality of device points and the previous point density parameter includes: determining a current device property of the device based on a current point displacement of each of the plurality of device points and the previous point density parameter; determining a current volume based on the previous point density parameters of each of the plurality of device points; Obtaining a point density reference parameter for each of the plurality of device points based on a derivative of a current device property of the device with respect to a point density parameter and a derivative of a current volume of the device with respect to the point density parameter; and The previous density field network loss value is determined based on the current device properties, the current volume, and the point density reference parameter.

7. The method according to claim 6, characterized in that The determining the previous density field network loss value based on the current device attribute, the current volume and the point density reference parameter includes: determining a first density loss value based on the current device properties and the initial properties of the device; determining a second density loss value based on the current volume and a target volume of the device; determining a third density loss value based on the previous dot density parameters of each of the plurality of device dots and the dot density reference parameters of each of the plurality of device dots; and The previous density field network loss value is determined based on the first density loss value, the second density loss value, and the third density loss value.

8. The method according to claim 6 or 7, characterized in that in, The density field network includes a plurality of cascaded fully connected layers; The adjusting parameters of the previous density field network based on the previous density field network loss value to obtain the current density field network includes: The parameters of the last fully connected layer in the previous density field network are adjusted based on the loss value of the previous density field network to obtain the current density field network.

9. The method according to claim 1, characterized in that The method further comprises: Using the current point density parameter as the target point density parameter; Obtaining a target Young's modulus of each of the plurality of device points based on the target point density parameter and the physical parameters of the target material to be filled in the device; and Based on the target Young's modulus of each of the plurality of device points, the target point displacement of each of the plurality of device points is determined to obtain the device structure optimization result.

10. An electronic device comprising: one or more processors; a memory for storing one or more computer programs, It is characterized in that the one or more processors execute the one or more computer programs to implement the steps of the method according to any one of claims 1 to 9.

Images