CN-121997772-A - Structure optimization design and debugging method and system based on image recognition

CN121997772ACN 121997772 ACN121997772 ACN 121997772ACN-121997772-A

Abstract

The invention discloses a structural optimization design and debugging method and system based on image recognition, which are characterized in that firstly, contour features are extracted from component images, a preliminary defect candidate region is positioned, further, the dimensional deviation in the region is subjected to fine classification and severity assessment through a neural network, geometric and material attribute deviation information is fused, a defect group distribution mode is analyzed through clustering, numerical simulation is triggered when the distribution density exceeds a limit to predict performance stability, and design and material parameters are reversely adjusted through an optimization algorithm according to stress and time domain response data obtained through simulation, so that a virtual assembly model and an optimization scheme which meet assembly precision standards are finally generated. The invention realizes the whole-flow closed loop from intelligent defect identification, group influence analysis to stability driving parameter optimization, and remarkably improves the quality control and design optimization efficiency of the complex mechanical device.

Inventors

KANG SHENG
ZHOU WEITING
YANG QIN
NIE JUNFEI

Assignees

邵阳学院

Dates

Publication Date: 20260508
Application Date: 20260401

Claims (10)

1. The structure optimization design and debugging method based on image recognition is characterized by comprising the following steps of: S100, acquiring component image data of a mechanical device, extracting outline features from the component image data through an image processing algorithm to obtain a preliminary defect candidate region, wherein the component image data comprises a plurality of part structures, and the preliminary defect candidate region corresponds to a potential abnormal part; S200, classifying the size deviation in the preliminary defect candidate region by adopting a neural network model, and determining a deviation type and a severity, wherein the deviation type relates to geometric and material properties, and the severity is classified based on the deviation degree; S300, acquiring assembly precision influence parameters from the deviation type and the severity, and grouping similar defects through a clustering method to obtain defect group distribution, wherein the defect group distribution represents a defect concentration mode; s400, judging whether the distribution density exceeds a threshold value according to the defect group distribution, and if so, simulating the performance stability by adopting a numerical simulation method to obtain a stability simulation result, wherein the stability simulation result comprises stress and time domain response data; S500, obtaining an optimization scheme generation basis from the stability simulation result, and adjusting design parameters through an optimization algorithm to determine parameter optimization values, wherein the parameter optimization values are adjusted according to the size and the materials; and S600, generating a virtual assembly model according to the parameter optimization value, judging whether the assembly precision in the virtual assembly model meets the standard, and outputting a final optimization scheme if the assembly precision meets the standard.
2. The method for optimizing design and debugging of an image recognition-based structure according to claim 1, wherein the step S100 comprises: s110, acquiring component image data of a mechanical device, and performing median filtering processing on the component image data to obtain smooth image data; S120, extracting contour features of the part structure from the smooth image data by utilizing an edge detection operator; s130, performing closed retrieval in a preset coordinate system according to the profile characteristics to obtain a closed profile set; and S140, if the geometric parameter deviation between the specific region in the closed contour set and the standard part template exceeds a preset threshold, determining the specific region as a preliminary defect candidate region.
3. The method for optimizing design and debugging of an image recognition-based structure according to claim 1, wherein step S200 comprises: S210, extracting multidimensional space geometrical features and gray level co-occurrence matrix texture features from the preliminary defect candidate region; S220, inputting the multidimensional space geometric features and the gray level co-occurrence matrix texture features into a depth residual error network model to obtain geometric attribute probability distribution and material attribute probability distribution; S230, determining the deviation type of the preliminary defect candidate region according to the geometrical attribute probability distribution and the material attribute probability distribution; S240, calculating a numerical deviation value if the deviation type belongs to geometric attribute deviation; s250, analyzing the texture heterogeneity if the deviation type belongs to the material attribute deviation; s260, determining severity according to the numerical deviation amount or the texture heterogeneity; s270, completing classification work of the size deviation in the preliminary defect candidate area through the deviation type and the severity.
4. The method for optimizing design and debugging of an image recognition-based structure according to claim 1, wherein the step S300 comprises: S310, searching assembly gaps and form and position tolerances in a preset mapping matrix according to the deviation type and the severity, and determining assembly accuracy influence parameters; s320, constructing a high-dimensional feature vector by utilizing the assembly precision influence parameters, and performing space attribute division on the preliminary defect candidate region by a density clustering algorithm to obtain a defect set with similar geometric features and material attributes; s330, calculating the fluctuation range of contact stress and material hardness in a local area aiming at the defect set, measuring the similarity between the defect sets through Euclidean distance, and determining a defect group; S340, extracting statistical characteristics of surface roughness and load distribution from the defect group, and calculating the spatial distribution density of the defects on the surface of the workpiece by using a nuclear density estimation method to obtain defect group distribution; S350, identifying an abnormal aggregation area of the thermal expansion quantity and the friction resistance according to the defect group distribution, and determining a defect concentration mode by analyzing the coupling relation between the fit tolerance and the structural rigidity.
5. The method for optimizing design and debugging of image recognition-based structure of claim 4, wherein step S400 comprises: S410, extracting local grid coordinates and intrinsic material properties from the defect group distribution to determine distribution density; s420, if the distribution density exceeds a preset critical evolution threshold, a finite element analysis model is called to obtain transient dynamics characteristics; S430, mapping a stiffness matrix and a damping matrix according to the transient dynamics characteristics, and obtaining a stability simulation result by iteratively solving a nonlinear motion equation; S440, carrying out component decomposition on the stress tensor in the stability simulation result to determine a stress distribution state; S450, matching the displacement vector with the acceleration signal by using the stress distribution state, and obtaining time domain response data through frequency spectrum transformation; S460, constructing a performance evaluation matrix through the stress distribution state and the time domain response data, and realizing quantitative evaluation of performance stability.
6. The method for optimizing design and debugging of an image recognition-based structure according to claim 1, wherein the step S500 comprises: s510, extracting a high-order feature vector from the performance evaluation matrix and mapping the high-order feature vector to a multidimensional design space to determine geometric topological variables to be corrected; s520, performing gradient descent iteration in a preset constraint range aiming at the geometric topological variable to acquire a size evolution sequence; s530, matching the corresponding composite component proportion according to the extreme points in the size evolution sequence, and searching a material attribute library to obtain a candidate enhancement phase proportion; S540, if the candidate enhancement matching ratio meets a preset structural strength criterion threshold, carrying out collaborative optimization on the geometric topological variable and the proportion of the composite component through a multi-target particle swarm algorithm to obtain an optimal configuration combination; s550, performing reconstruction processing on the original design model through the optimal configuration combination, and determining parameter optimization values adjusted for the size and the material.
7. The method for optimizing design and debugging of an image recognition-based structure according to claim 1, wherein the step S600 comprises: S610, constructing a geometric solid model in a three-dimensional modeling engine according to the parameter optimization value, wherein the geometric solid model is endowed with corresponding physical attribute materials; The geometric solid model is constructed by the following formula: ; Wherein, the The geometric solid model is represented by a model of the geometric solid, Representing the three-dimensional modeling engine, Representing a set of vertices of a set of points, A set of facets is represented and, Representing the build parameters; S620, performing space pose alignment according to the geometric entity model in a virtual simulation environment according to a preset topological connection relationship so as to generate a virtual assembly model; s630, carrying out grid division on the contact surface in the virtual assembly model, extracting normal vector deviation of the matching surface, and judging whether the normal vector deviation is in a preset tolerance fluctuation interval or not; s640, if the normal vector deviation is in a preset tolerance fluctuation interval, extracting mass center offset among all components and calculating assembly precision; S650, judging whether the assembly precision is smaller than or equal to a preset error threshold, and outputting a final optimization scheme if the assembly precision is smaller than or equal to the preset error threshold.
8. The method for optimizing design and debugging of image recognition based structure according to claim 7, wherein in step S620, the virtual assembly model is generated by the following formula: ; Wherein, the Representing a virtual assembly model of the vehicle, Representing a topological join operation in a virtual simulation environment, Representing the aligned first The geometric solid model is used for generating a geometric solid model, Representing the total number of geometric entities; Aligned first Geometric solid model The method is characterized by comprising the following steps: ; Wherein, the Representing the aligned first The geometric solid model is used for generating a geometric solid model, Representing a topology-based pose transformation function, Representing original geometric entity The geometric solid model is used for generating a geometric solid model, Represent the first Topological neighborhood of the geometric entities.
9. The method for optimizing design and debugging of image recognition based structure of claim 8, wherein in step S630, the normal vector deviation is obtained by the following formula: ; Wherein, the Representing the deviation of the normal vector and, Representing the normal vector of the contact surface, Representing the normal vector of the mating surface.
10. A structural optimization design and debugging system based on image recognition, for executing the structural optimization design and debugging method based on image recognition as claimed in any one of claims 1 to 9, comprising: The device comprises a preliminary defect candidate region acquisition module (10) for acquiring component image data of a mechanical device, extracting outline features from the component image data through an image processing algorithm to obtain a preliminary defect candidate region, wherein the component image data comprises a plurality of part structures, and the preliminary defect candidate region corresponds to a potential abnormal part; a deviation type and severity determination module (20) for classifying dimensional deviations within the preliminary defect candidate areas using a neural network model, determining a deviation type and a severity, wherein the deviation type relates to geometric and material properties, the severity being graded based on the degree of deviation; a defect group distribution acquisition module (30) for acquiring assembly precision influence parameters from the deviation type and the severity, grouping similar defects through a clustering method, and obtaining defect group distribution, wherein the defect group distribution represents a defect concentration mode; A stability simulation result obtaining module (40) for judging whether the distribution density exceeds a threshold value according to the defect group distribution, and if so, simulating the performance stability by adopting a numerical simulation method to obtain a stability simulation result, wherein the stability simulation result comprises stress and time domain response data; The parameter optimization value determining module (50) is used for obtaining an optimization scheme generation basis from the stability simulation result, adjusting design parameters through an optimization algorithm, and determining a parameter optimization value, wherein the parameter optimization value is adjusted for the size and the material; and the final optimization scheme output module (60) is used for generating a virtual assembly model according to the parameter optimization value, judging whether the assembly precision in the virtual assembly model meets the standard or not, and outputting the final optimization scheme if the assembly precision meets the standard.

Description

Structure optimization design and debugging method and system based on image recognition Technical Field The invention relates to the technical field of industrial manufacturing and mechanical design, and particularly discloses a structure optimization design and debugging method and system based on image recognition. Background In the field of industrial manufacturing and mechanical design, structural optimization and debugging are important links for ensuring equipment performance and production efficiency, and the process directly relates to the quality of products and the core competitiveness of enterprises. Along with the continuous improvement of the complexity of industrial equipment, how to improve the design precision and the debugging efficiency by technical means becomes an urgent requirement for industry development. Traditional structural design and debugging methods often depend on manual experience, are difficult to cope with the requirements of modern industry on high precision and high efficiency, and need to introduce intelligent means to break through the existing bottleneck. Currently, most solutions, when faced with complex mechanical structures, often suffer from a lack of dynamic capture and comprehensive analysis capabilities of the structural details, resulting in design deviations and debugging problems that are difficult to discover and resolve in time. Especially under the condition of multi-part assembly and running environment change, the existing method often cannot accurately identify fine defects, and systematic adjustment and optimization of design parameters are difficult. The limitation makes the design and debugging process time-consuming and high in cost, and influences the whole research and development period of industrial equipment. A further technical challenge is that there is a tightly linked difficulty between the identification of structural defects and the generation of an optimization scheme. The first key factor is the accurate capturing and analysis of structural images, the shapes of parts of industrial equipment are different, the sizes are complex, and if contour and size information cannot be accurately extracted from the images, hidden defects or deviations are difficult to find. Next, this problem further affects the formulation of the optimization scheme, because the lack of accurate deviation information, the direction and magnitude of the design adjustment cannot be scientifically and reasonably designed, resulting in that the optimization scheme may deviate from the actual requirements. For example, in the design of a liquid dispenser, if the dimensional deviation of the connection position of a certain key part is not accurately identified, the optimization scheme may not effectively improve the assembly precision, and finally the stability of the operation of the device is affected. Therefore, how to accurately identify defects and generate a scientific and reasonable optimization scheme through high-efficiency analysis of structural images in a complex industrial scene becomes a key problem for improving design and debugging efficiency. Disclosure of Invention The invention provides a structure optimization design and debugging method and system based on image recognition, and aims to accurately recognize defects and generate a scientific and reasonable optimization scheme through high-efficiency analysis of a structure image. One aspect of the invention relates to a structure optimization design and debugging method based on image recognition, which comprises the following steps: s100, acquiring component image data of a mechanical device, extracting contour features from the component image data through an image processing algorithm to obtain a preliminary defect candidate region, wherein the component image data comprises a plurality of part structures, and the preliminary defect candidate region corresponds to a potential abnormal part; s200, classifying the size deviation in the preliminary defect candidate area by adopting a neural network model, and determining the deviation type and the severity, wherein the deviation type relates to geometric and material properties, and the severity is classified based on the deviation degree; S300, acquiring assembly precision influence parameters from deviation types and severity, and grouping similar defects through a clustering method to obtain defect group distribution, wherein the defect group distribution represents a defect concentration mode; S400, judging whether the distribution density exceeds a threshold value according to the defect group distribution, and if so, simulating the performance stability by adopting a numerical simulation method to obtain a stability simulation result, wherein the stability simulation result comprises stress and time domain response data; S500, obtaining an optimization scheme generation basis from a stability simulation result, adjusting design parameters thr