Search

CN-122021266-A - Cement-based 3D printing collapse resistance evaluation method and system based on data driving

CN122021266ACN 122021266 ACN122021266 ACN 122021266ACN-122021266-A

Abstract

The invention provides a data-driven-based cement-based 3D printing collapse resistance assessment method and system; the method comprises the steps of constructing a data set through a material property test and finite element simulation, wherein the data set comprises input features of each printing layer and corresponding constructability state classification labels, the input features comprise normalized geometric descriptors extracted from special-shaped components, environment temperature and material age, a multi-classification model of a decision tree is lifted by utilizing a data set training gradient so as to learn complex mapping between the features and the labels, the trained model is integrated into 3D printing slicing software, the target components are sliced layer by layer and extracted in the printing preparation stage, the model is called in real time to predict and early warn, the obvious improvement of evaluation efficiency is realized, plastic collapse and buckling instability can be predicted in real time at the same time, and the special-shaped components can be generally evaluated through the normalization features.

Inventors

  • WANG XINGYU
  • MA MINGLEI
  • YANG YAN
  • HAN LIFANG
  • FENG MINGYANG
  • BAI JIE

Assignees

  • 中国建筑第八工程局有限公司

Dates

Publication Date
20260512
Application Date
20260112

Claims (10)

  1. 1. The method for evaluating the collapse resistance of the cement-based 3D printing based on data driving is characterized by comprising the following steps of: s1, acquiring a data set for training a machine learning model, wherein the data set comprises input features of each printing layer and corresponding constructable state classification labels in a 3D printing process; s2, training a machine learning multi-classification model by utilizing the data set to establish a mapping relation from the input features to the classification labels; S3, integrating the machine learning multi-classification model after training into an evaluation system; S4, slicing the three-dimensional model of the target printing component in a printing preparation stage, and extracting input features of a current layer by layer; s5, inputting the extracted input features of the current layer into the machine learning multi-classification model, and predicting and outputting the constructable state classification labels of the current layer in real time so as to realize collapse resistance assessment and early warning in the printing process.
  2. 2. The method for evaluating the collapse resistance of data-driven-based 3D printing according to claim 1, wherein in the step S1, the construction of the data set comprises the following steps: S11, calibrating time-varying mechanical property parameters of cement-based materials along with time and environmental temperature changes based on a material property test; S12, extracting multidimensional features including geometric features, ambient temperature and material age as the input features for each printing layer in the 3D printing process; and S13, labeling the constructable state classification labels for each printing layer based on finite element simulation or experiment.
  3. 3. The method for evaluating the collapse resistance of the data-driven-based 3D printing according to claim 2, wherein in the step S11, a material property function of the coupling temperature and the time effect is constructed by performing material property tests under different temperature gradients to obtain a change curve of yield strength and elastic modulus along with the age of the material.
  4. 4. The data-driven based 3D printing collapse resistance assessment method according to claim 2, wherein in step S12, the geometric features include line features, local features, cross-sectional features, and overall features; the line features include line eccentricity; The local features include at least one of local tilt angles and relative heights; The cross-sectional features include at least one of a cross-sectional fourier descriptor, a convex hull area ratio, and a shape complexity factor; The global features include at least one of relative height ratios and cumulative bias.
  5. 5. The method for evaluating the collapse resistance of the data-driven cement-based 3D printing according to claim 4, wherein the cross-section Fourier descriptor is obtained by performing discrete Fourier transform on a cross-section contour coordinate sequence of the printing layer and extracting low-frequency components of the cross-section Fourier descriptor, wherein the low-frequency components are used for describing macroscopic shape characteristics of a cross section.
  6. 6. The method for evaluating the collapse resistance of data-driven cement-based 3D printing according to claim 2, wherein in step S13, the constructable status classification labels are labeled for each printing layer based on finite element simulation or experiment, comprising the steps of: s131, generating a plurality of special-shaped component models with different geometric topological characteristics through parameterized modeling; s132, simulating the printing stacking process of each component layer by utilizing finite element simulation and adopting a life-death unit technology, and calling the time-varying mechanical property parameters; And S133, automatically labeling the classification labels for each printing layer according to the simulation result.
  7. 7. The method for evaluating the collapse resistance of a data-driven based 3D printing system according to claim 2, wherein in step S12, the extracted geometric features are calculated as normalized dimensionless parameters.
  8. 8. The method for evaluating the collapse resistance of the data-driven based 3D printing system according to claim 1, wherein in the step S2, the machine learning multi-classification model is a gradient lifting decision tree model.
  9. 9. The method for evaluating the collapse resistance of a data-driven based 3D printing system according to claim 1, wherein in step S3, the evaluation system is integrated into 3D printing slicing software for real-time evaluation during generation of slicing paths.
  10. 10. A data-driven based 3D printing collapse resistance assessment system, comprising: a feature extraction module for performing a layer-by-layer analysis of the target printing member during a print preparation or execution phase, extracting input features of the current layer, the input features being defined according to the method of any one of claims 1 to 9; A model evaluation module storing a machine-learned multi-classification model trained by the method of any of claims 1-9 for receiving the input features and outputting a current layer of constructable state classification labels; And the early warning module is used for generating early warning information according to the constructability state classification label.

Description

Cement-based 3D printing collapse resistance evaluation method and system based on data driving Technical Field The invention belongs to the technical field of building 3D printing, and particularly relates to a data-driven cement-based 3D printing collapse resistance assessment method and system. Background 3D concrete printing technology (3 DCP) has spanned from simple component fabrication to the automated construction phase of an entire house since the beginning of the 21 st century. Early studies focused on "printability" itself, i.e., whether the material was successfully extruded. However, as the demand for complex shaped, non-linear curved surface forms for building aesthetics increases, the assessment of constructability during construction becomes a technical core. In traditional civil engineering, the stability of a structure is often statically checked based on a fixed design strength. However, during 3D printing, the structure is always in a dynamic load state of "growing while hardening". In recent years, the academia tries to introduce sensor monitoring and finite element simulation, but the former lacks prejudgement, and the latter is difficult to adapt to the requirement of real-time feedback of a printing field due to high time cost of three-dimensional nonlinear calculation. Cement-based 3D printing has significant advantages in building shaped structures, but its construction process has extremely high uncertainty. First, cement-based materials such as concrete have significant time-varying properties, and their strength development is severely affected by environmental factors, resulting in unpredictable collapse risks at the construction site. Second, the profiled structure (e.g., torsion, cantilever) can create complex additional bending moments and shear stresses. The existing finite element analysis method can simulate the process, but has extremely high calculation cost, and single simulation usually takes several hours, so that the real-time evaluation requirement of quick response of a printing field cannot be met. In addition, the industry lacks a general stability assessment framework that can uniformly describe different geometries and can compromise environmental factors and printing processes. Disclosure of Invention Aiming at the defects existing in the prior art, the invention provides a data-driven-based method and a data-driven-based system for evaluating the collapse resistance of the 3D printing, which can simultaneously consider environmental factors and printing processes, are suitable for uniform stability evaluation of various special-shaped geometries, shorten evaluation time and realize real-time early warning of collapse and instability risks. The technical scheme of the invention is that the data-driven-based cement-based 3D printing collapse resistance assessment method comprises the following steps: s1, acquiring a data set for training a machine learning model, wherein the data set comprises input features of each printing layer and corresponding constructable state classification labels in a 3D printing process; s2, training a machine learning multi-classification model by utilizing the data set to establish a mapping relation from the input features to the classification labels; S3, integrating the machine learning multi-classification model after training into an evaluation system; S4, slicing the three-dimensional model of the target printing component in a printing preparation stage, and extracting input features of a current layer by layer; s5, inputting the extracted input features of the current layer into the machine learning multi-classification model, and predicting and outputting the constructable state classification labels of the current layer in real time so as to realize collapse resistance assessment and early warning in the printing process. Further, in step S1, the construction of the data set includes the following steps: S11, calibrating time-varying mechanical property parameters of cement-based materials along with time and environmental temperature changes based on a material property test; S12, extracting multidimensional features including geometric features, ambient temperature and material age as the input features for each printing layer in the 3D printing process; and S13, labeling the constructable state classification labels for each printing layer based on finite element simulation or experiment. Further, in step S11, a change curve of yield strength and elastic modulus with age of the material is obtained by performing a material property test under different temperature gradients, and a material property function of coupling temperature and time effect is constructed. Further, in step S12, the geometric features include line features, local features, cross-sectional features, and global features; the line feature includes line eccentricity ; The local features include local dip anglesAnd relative height ofAt least one of (a