CN-121972793-A - Laser three-dimensional intelligent hole making method and system for air film holes

CN121972793ACN 121972793 ACN121972793 ACN 121972793ACN-121972793-A

Abstract

The invention provides a method and a system for three-dimensional intelligent hole making of a gas film hole laser, which comprise the steps of constructing a multi-layer material digital twin model, establishing a laser-material interaction physical model, converting the model into an edge value problem of an energy conservation system, obtaining a three-dimensional energy distribution field by numerical solution, realizing processing parameter optimization by combining deep reinforcement learning, and generating an optimal processing track. The invention obviously improves the machining precision and consistency of the air film holes, reduces the defect formation and provides an advanced solution for the manufacture of the hot end parts of the aero-engine.

Inventors

YAO WEI
GUO ZHIRUI
FANG XIAOYU

Assignees

贵州航谷动力科技有限公司

Dates

Publication Date: 20260505
Application Date: 20260408

Claims (11)

1. The method for laser three-dimensional intelligent hole making of the air film hole is characterized by comprising the following steps of: acquiring multi-dimensional characteristic data of a workpiece, preprocessing the multi-dimensional characteristic data, and generating a standardized data set; based on the standardized data set, a workpiece state model is established by adopting a random time-lag model, and a digital twin model is established by combining a numerical analysis and a deep learning method to obtain digital twin model parameters; Based on the digital twin model parameters, a physical model of interaction between laser and materials is established, a laser-material interaction problem is converted into an edge value problem of an energy conservation system, a numerical value solver is adopted to solve the edge value problem, a three-dimensional energy distribution field is output, a mapping relation between processing parameters and processing quality is established, an intelligent optimization algorithm is adopted to solve a parameter optimization problem, and a dynamic parameter adjustment strategy and an optimal processing track are generated; And constructing a multistage closed-loop control system based on the dynamic parameter adjustment strategy and the optimal processing track, acquiring processing process information in real time through a sensor to obtain real-time feedback information, inputting the real-time feedback information into the digital twin model to perform simulation calculation to obtain a prediction result of the development trend of the processing process, and dynamically adjusting laser processing parameters and the processing track according to the prediction result and the real-time feedback information to finish processing of the air film hole.
2. The method of claim 1, wherein the multi-dimensional characterization data comprises geometry data, internal structure data, material energy absorption characteristic data, thermal barrier coating thickness data, and surface roughness data.
3. The method of claim 2, wherein preprocessing the multi-dimensional feature data to generate a normalized dataset comprises: The method comprises the steps of obtaining geometric shape data of the outer surface of a workpiece through a high-precision three-dimensional laser scanner, obtaining internal structure data of the workpiece through X-ray computed tomography, and acquiring material energy absorption characteristic data, thermal barrier coating thickness data and surface roughness data of the workpiece through a thermal imager and an ultrasonic detector to obtain multi-source heterogeneous data; And filtering the multi-source heterogeneous data, removing noise interference, registering the filtered data, unifying the coordinate system of the multi-source heterogeneous data, and fusing the registered data to generate a standardized data set.
4. The method of claim 3, wherein the step of establishing a workpiece state model by using a random time-lag model, and establishing a digital twin model by combining a numerical analysis and a deep learning method to obtain digital twin model parameters comprises the steps of: Extracting geometric parameters, material characteristic parameters and thermal barrier coating parameters of a workpiece from the standardized dataset as initial state variables, constructing a workpiece state vector based on the initial state variables, and representing the workpiece state vector as a random time-lag differential equation with Markov switching characteristics, wherein the workpiece state vector is jointly influenced by a time-lag function, a Markov chain and a random process, the Markov chain represents random switching of a processing environment, and a workpiece state model is established; Correlating the workpiece state model with historical processing data, learning a mapping relation between the workpiece state vector and a processing response by adopting a deep neural network, and establishing a temperature field evolution model, a stress field evolution model and a material removal model of the workpiece in the laser processing process by combining a numerical analysis method to construct a digital twin model; And training the deep neural network to obtain network weight, bias parameters and model super parameters, and determining a heat conduction coefficient of a temperature field evolution model, an elastic modulus of a stress field evolution model and a removal rate coefficient of a material removal model by combining a numerical analysis method and the initial state variable to obtain the digital twin model parameters.
5. The method of claim 4, further comprising, after the obtaining the digital twin model parameters: in the laser processing process, acquiring temperature distribution data, stress strain data and material removal depth data of a workpiece in real time through a deployed sensor to obtain sensing data, and converting the sensing data into a real-time state vector with the same dimension as the state vector of the workpiece; Comparing the real-time state vector with the workpiece state vector, and judging that the workpiece state changes when the deviation of the real-time state vector and the workpiece state vector exceeds a preset threshold value, so as to trigger a model updating flow; And taking the sensing data, the corresponding processing parameters and the processing quality evaluation result as newly-increased historical processing experience, combining the newly-increased historical processing experience with the historical processing data to form an updated historical processing data set, continuously optimizing the digital twin model parameters from the historical processing data set by adopting an incremental learning algorithm, and ensuring the consistency of the digital twin model and the actual workpiece state.
6. The method of claim 5, wherein the modeling of the laser-material interaction to transform the laser-material interaction problem into a side value problem for the energy conservation system comprises: Based on the digital twin model parameters, combining the optical characteristics, thermal characteristics and mechanical characteristics of the materials, establishing a physical model of the interaction of laser and the multi-layer materials, wherein the physical model comprises a light absorption process, a heat conduction process, a phase change process and a plasma forming and shielding process, and forming a partial differential equation set for describing energy transfer; And converting the partial differential equation set into a boundary value problem of the Hamiltonian system based on an energy conservation principle, and establishing complete boundary value problem description aiming at different material interfaces and boundary conditions.
7. The method of claim 6, wherein said solving the edge problem with a numerical solver outputs a three-dimensional energy distribution field, comprising: Discretizing the complete side value problem description, dividing a solving domain into grid units by adopting a finite element method, and converting the side value problem of the Hamiltonian system into a large-scale linear equation set; Solving the large-scale linear equation set by adopting a precondition conjugate gradient method, obtaining the temperature field distribution, stress field distribution and energy density distribution of each grid node through iterative calculation, and outputting a three-dimensional energy distribution field.
8. The method of claim 7, wherein solving the parameter optimization problem using the intelligent optimization algorithm comprises: Based on the three-dimensional energy distribution field, a mapping relation between laser power, frequency, pulse width and focusing position, machining precision, surface roughness and a heat affected zone is established, a random time-lag differential equation is adopted to describe the influence of the changes of the laser power, frequency, pulse width and focusing position on the machining precision, surface roughness and the heat affected zone, a machining parameter optimization problem is established as a Markov decision process, a state space is the current machining state of a workpiece, an action space is an adjustable parameter set of the laser power, frequency, pulse width and focusing position, a reward function is designed based on evaluation indexes of the machining precision, surface roughness and the heat affected zone, state transition probability is calculated by combining the digital twin model parameters and the three-dimensional energy distribution field, the Markov decision process is solved by adopting a depth reinforcement learning algorithm, and dynamic parameter adjustment strategies aiming at different workpiece machining states are generated.
9. The method of claim 8, wherein generating an optimal machining trajectory comprises: Based on the dynamic parameter adjustment strategy, discretizing a target hole pattern into grid representation, generating a processing point set, taking space coordinates, energy requirements and time sequence constraint of each processing point as system variables, constructing a system matrix according to space adjacency relations, energy transfer relations and time sequence dependency relations among the processing points, wherein elements of the system matrix represent association strength among the processing points, establishing a complex symmetric linear system of track planning based on the system matrix, decomposing the system matrix, solving the complex symmetric linear system by adopting an iterative method and combining a preprocessor, and generating an optimal processing track comprising a laser moving path, residence time and an incidence angle; Based on the dynamic parameter adjustment strategy, comparing the real-time state vector obtained by converting the real-time feedback information with an expected state corresponding to a preset track in the processing process to be used as real-time processing feedback, establishing a track dynamic adjustment mechanism, triggering a track re-planning process when the real-time processing feedback shows material characteristic change or plasma shielding enhancement, reconstructing the system matrix, solving the complex symmetrical linear system, and adjusting the optimal processing track in real time.
10. The method according to claim 9, wherein the constructing the multi-stage closed-loop control system, acquiring the processing information in real time through a sensor to obtain real-time feedback information, inputting the real-time feedback information into the digital twin model to perform simulation calculation to obtain a prediction result of the development trend of the processing, dynamically adjusting the laser processing parameters and the processing track according to the prediction result and the real-time feedback information, and completing the processing of the gas film hole, comprises: Based on the dynamic parameter adjustment strategy and the optimal processing track, a three-level closed-loop control system comprising inner ring parameter regulation, middle ring track adjustment and outer ring quality optimization is constructed, a high-speed camera, a spectrometer and an acoustic emission sensor are deployed to collect processing process information in real time, and a multi-source data monitoring hole pattern forming process, a material removal state and processing defects are integrated through a sensor fusion algorithm to obtain real-time feedback information; Inputting the real-time feedback information into the digital twin model, performing simulation calculation by using the digital twin model, predicting hole pattern evolution, temperature field distribution and material removal rate at future time to obtain a prediction result of the development trend of the processing process, and triggering an early warning mechanism when the prediction result exceeds a preset safety range; The method comprises the steps of carrying out real-time adjustment on laser power, pulse width, frequency and focusing depth according to real-time feedback information and a prediction result based on inner ring parameter adjustment, carrying out real-time adjustment on a laser moving path according to machining state abnormality displayed by the prediction result based on middle ring track adjustment, carrying out on-line detection and evaluation on a machined air film hole based on outer ring quality optimization, storing machining parameters, process data and quality evaluation results into a knowledge base, excavating parameter-quality relation through a machine learning method, and continuously optimizing a machining strategy to finish machining the air film hole.
11. Air film hole laser three-dimensional intelligence system of making hole, its characterized in that includes: The data acquisition and preprocessing module is used for acquiring the multi-dimensional characteristic data of the workpiece, preprocessing the multi-dimensional characteristic data and generating a standardized data set; The digital twin modeling module is used for establishing a workpiece state model by adopting a random time-lag model based on the standardized data set, and establishing a digital twin model by combining a numerical analysis and a deep learning method to obtain digital twin model parameters; The physical model solving and optimizing module is used for establishing a physical model of interaction between laser and materials based on the digital twin model parameters, converting the laser-material interaction problem into an edge value problem of an energy conservation system, adopting a numerical solver to solve the edge value problem, outputting a three-dimensional energy distribution field, establishing a mapping relation between processing parameters and processing quality, adopting an intelligent optimization algorithm to solve a parameter optimization problem, and generating a dynamic parameter adjustment strategy and an optimal processing track; The multistage closed-loop control module is used for constructing a multistage closed-loop control system based on the dynamic parameter adjustment strategy and the optimal processing track, acquiring processing process information in real time through a sensor to obtain real-time feedback information, inputting the real-time feedback information into the digital twin model to perform simulation calculation to obtain a prediction result of the development trend of the processing process, and dynamically adjusting laser processing parameters and the processing track according to the prediction result and the real-time feedback information to finish processing of the air film hole.

Description

Laser three-dimensional intelligent hole making method and system for air film holes Technical Field The invention relates to the technical field of laser processing, in particular to a method and a system for laser three-dimensional intelligent hole making of a gas film hole, which are used for processing gas film cooling holes of high-temperature components such as a flame tube, a nozzle and the like of a combustion chamber of an aeroengine. Background High-temperature components such as a flame tube, a nozzle and the like of a combustion chamber of an aeroengine need a large number of air film cooling holes so as to ensure the reliability and the service life of the high-temperature components in an extreme working environment. The shape, size, distribution and angle of these cooling holes have a decisive influence on the cooling effect, and therefore place extremely high demands on the machining accuracy and consistency. Traditional air film hole processing mainly adopts electric spark processing or mechanical drilling technology. The electric spark machining can realize the machining of holes with complex shapes, but has low machining efficiency and is easy to cause a surface heat affected zone, and the mechanical drilling has larger limitation in the machining of non-straight holes and is easy to generate defects of coating peeling, layering and the like when facing to composite materials such as thermal barrier coatings and the like. Laser processing is becoming an important method for processing gas film holes due to the advantages of non-contact property, high precision and the like. The existing laser three-dimensional hole-making technology generally adopts a preset parameter open-loop processing mode, sets parameters such as laser power, pulse width and the like based on experience, and then completes processing according to a preset track. The mode can not be adjusted in real time according to factors such as material property change, heat accumulation effect, plasma shielding formed in the processing process and the like in the actual processing process, so that the problems of large hole shape precision difference, difficult processing depth control, poor batch consistency and the like are caused. In addition, in the prior art, when facing complex components of an aeroengine such as a Y-shaped hole, a cat-ear shaped hole and other abnormal cooling structures and carbon-based Composite Materials (CMC), energy absorption and material removal behavior change in the processing process cannot be effectively predicted and compensated, high-precision closed-loop control is difficult to realize, unstable processing quality is caused, and component performance and service life are affected. Disclosure of Invention The invention provides a method and a system for three-dimensional intelligent hole making of a gas film hole laser, and aims to solve the technical problems that the existing laser three-dimensional hole making technology cannot adjust in real time aiming at factors such as material property change, heat accumulation effect, plasma shielding formed in the processing process and the like, so that the hole shape precision difference is large, the processing depth is difficult to control, the batch consistency is poor and the like. In order to achieve the above purpose, the invention provides a laser three-dimensional intelligent hole making method for a gas film hole, comprising the following steps: acquiring multi-dimensional characteristic data of a workpiece, preprocessing the multi-dimensional characteristic data, and generating a standardized data set; based on the standardized data set, a workpiece state model is established by adopting a random time-lag model, and a digital twin model is established by combining a numerical analysis and a deep learning method to obtain digital twin model parameters; Based on the digital twin model parameters, a physical model of interaction between laser and materials is established, a laser-material interaction problem is converted into an edge value problem of an energy conservation system, a numerical value solver is adopted to solve the edge value problem, a three-dimensional energy distribution field is output, a mapping relation between processing parameters and processing quality is established, an intelligent optimization algorithm is adopted to solve a parameter optimization problem, and a dynamic parameter adjustment strategy and an optimal processing track are generated; And constructing a multistage closed-loop control system based on the dynamic parameter adjustment strategy and the optimal processing track, acquiring processing process information in real time through a sensor to obtain real-time feedback information, inputting the real-time feedback information into the digital twin model to perform simulation calculation to obtain a prediction result of the development trend of the processing process, and dynamically adjusting laser pro