Intelligent design method for SiC fiber reinforced metal matrix composite pipe structure based on data driving

Abstract

The invention relates to the field of composite material structure design and manufacturing, in particular to an intelligent design method for a SiC fiber reinforced metal matrix composite material pipe structure based on data driving. Firstly, to-be-designed structure parameters in a pipe structure are defined as design variables; then performing test design, and obtaining a plurality of structural parameter combinations to be subjected to analogue simulation based on a test design algorithm; then data acquisition is carried out, analogue simulation is carried out on structure parameter combinations obtained through test design based on a finite element calculation method, and the service performance of the pipe structure corresponding to each structure parameter combination is obtained; carrying out model training; solving based on an optimization algorithm to obtain an optimal structure parameter combination; and finally, result verification is carried out, a corresponding SiC fiber reinforced metal matrix composite pipe structure is prepared, a service performance test is carried out, and a design result is verified. According to the method, the development period of the composite tubular structural member can be effectively shortened, and the service performance of the composite tubular structural member is improved.

Description

Technical Field

[0001] The present invention relates to the field of composite material structure design and manufacturing, and in particular to a data-driven intelligent design method for a SiC fiber reinforced metal matrix composite material pipe structure. Background Art

[0002] SiC fiber reinforced metal matrix composite is a composite material with high strength, high modulus, low density SiC fiber as reinforcement and metal material as matrix. SiC fiber reinforced metal matrix composite has attracted wide attention in the aerospace field due to its high specific strength, high specific stiffness, excellent creep resistance and high temperature durability.

[0003] Tubular structural parts are common load-bearing structural parts in aerospace vehicles. Compared with metal materials, the use of SiC fiber reinforced metal matrix composite materials to prepare tubular structural parts can significantly reduce the weight of structural parts while maintaining the original service performance, which is of great significance to the lightweighting of aircraft. At present, the structure of SiC fiber reinforced metal matrix composite tubular structural parts is relatively complex, involving many structural parameters, and the design is very difficult. However, the production cost of this type of material is high and the component preparation cycle is long. The experimental cost and development cycle of the traditional trial and error design method cannot be met at this stage. Summary of the invention

[0004] In view of the limitations of the prior art, the purpose of the present invention is to propose a data-driven intelligent design method for SiC fiber reinforced metal matrix composite tube structures, which can realize the intelligent design of SiC fiber reinforced metal matrix composite tubular structures with high efficiency and high precision under the drive of finite element simulation data, shorten the R&D design cycle, and improve the R&D design efficiency.

[0005] The present invention is achieved through the following technical solutions:

[0006] A data-driven intelligent design method for SiC fiber reinforced metal matrix composite pipe structure includes the following steps:

[0007] Step 1 is problem definition, which defines the structural parameters to be designed in the pipe structure as design variables, and divides the weight of the pipe structure and the service performance under different working conditions into design objectives and design constraints according to requirements;

[0008] Step 2 is experimental design, which obtains multiple structural parameter combinations to be simulated based on the experimental design algorithm;

[0009] Step 3 is data acquisition. Based on the finite element calculation method, the structural parameter combinations obtained by the experimental design are simulated to obtain the service performance of the pipe structure corresponding to each structural parameter combination;

[0010] Step 4 is model training, which uses the structural parameters as data features and the service performance of the corresponding pipe structure as data labels to train a machine learning model for predicting the service performance of the pipe structure;

[0011] Step 5 is optimization solution, which is solved based on the optimization algorithm. During the solution process, the model trained in step 4 is used to calculate the required service performance results, so as to obtain the optimal combination of structural parameters;

[0012] Step 6 is result verification. The SiC fiber reinforced metal matrix composite tube structure corresponding to the structural parameter combination obtained in step 5 is prepared and a service performance test is carried out. If it meets the design index requirements, the final design result is obtained; if it does not meet the design index requirements, steps 2 to 5 are repeated to increase the data volume and update the machine learning model until the final design result is obtained.

[0013] The data-driven intelligent design method for SiC fiber reinforced metal matrix composite tube structure is a tubular structure formed by coaxially compounding an inner metal matrix, a composite material layer, and an outer metal matrix in sequence. The outer shape of the tube structure is a cylinder of equal outer diameter. The SiC fiber reinforced metal matrix composite material layer is arranged in a metal matrix tube wall composed of an inner metal matrix and an outer metal matrix. In the composite material layer, SiC fibers with metal coatings deposited on the surface are arranged in the metal matrix tube wall as reinforcements, and the two ends of the composite material layer and the tube structure joint are arranged in a stepped manner. The tube structure is divided into a central section and a joint section. The joint section is a portion where the two ends of the composite material layer are connected to the tube structure. The central section is a non-joint section portion of the SiC fiber reinforced metal matrix composite tube structure where SiC fibers are uniformly arranged. The wall thickness of the tube structure joint section is greater than the wall thickness of the central section.

[0014] The data-driven intelligent design method for SiC fiber reinforced metal matrix composite tube structure, the problem definition of step 1 refers to: defining the design problem of SiC fiber reinforced metal matrix composite tube structure, including the following steps: 1) defining the structural parameters of the tube structure to be designed and the value range of the parameters, including but not limited to the inner metal matrix thickness h in , the thickness of the outer metal substrate h out , pipe structure joint section increases wall thickness h head , thickness of composite material layer h c , the number of fiber layers in the composite material layer N c , the length difference of each fiber layer in the composite material layer is Δd c , the distance d between the longest fiber layer in the composite material layer and the two ends of the tube structure head ; 2) define the design objectives and design constraints of the pipe structure, the variables involved are: pipe structure weight G, maximum load F that the pipe structure can withstand under axial tension s, the maximum load F that the pipe structure can withstand under axial compression conditions c , these variables are divided into design objectives and design constraints according to the usage scenarios of the pipe structure, and the boundary values ​​of the design constraints are defined.

[0015] In the data-driven intelligent design method for SiC fiber reinforced metal matrix composite tube structure, the experimental design in step 2 refers to: determining the amount of data set required for building the model based on the defined SiC fiber reinforced metal matrix composite tube structure design problem, and determining the structural parameter combination of the data based on the experimental design algorithm, the algorithm includes but is not limited to full factor sampling, orthogonal experimental design, Latin hypercube sampling or optimal Latin hypercube sampling.

[0016] In the data-driven intelligent design method for SiC fiber reinforced metal matrix composite tube structure, data acquisition in step 3 refers to: based on the finite element simulation calculation method, calculating the service performance of the tube structure corresponding to each structural parameter combination in the data set, specifically including but not limited to the following steps: tube structure geometric model construction, finite element simulation pre-processing model construction, finite element simulation model calculation and solution, finite element simulation result post-processing; then, using the structural parameter combination as the data feature and the corresponding service performance as the data label to form the data set required for modeling.

[0017] In the data-driven intelligent design method for SiC fiber reinforced metal matrix composite tube structure, the model training in step 4 refers to: based on a machine learning method, using the data set established in the data acquisition step, training a machine learning model for predicting the service performance of the tube structure, the machine learning method includes but is not limited to support vector regression, neural network, Kriging model, gradient boosting tree or random forest.

[0018] In the data-driven intelligent design method for SiC fiber reinforced metal matrix composite tube structure, the optimization solution in step 5 refers to: solving the problem defined in the problem definition step based on a global optimization algorithm, the optimization algorithm includes but is not limited to a genetic algorithm, a differential algorithm or a particle swarm algorithm, and using a machine learning model to calculate the values ​​of the design objectives and design constraints during the solution process, and multiple rounds of iterations are performed to obtain a structural parameter combination that optimizes the design objectives and satisfies all design constraint boundary values.

[0019] The design concept of the present invention is:

[0020] The present invention introduces a data-driven method to replace the traditional trial-and-error design method: the structural parameters of the SiC fiber-reinforced metal matrix composite material tube are designed; the finite element simulation method is used to replace the experimental data acquisition with high preparation cost and long test cycle; machine learning is used to establish a high-precision, low-computational-time service performance prediction model; and a global optimization algorithm is used to search for the optimal combination of structural parameters.

[0021] The core idea of ​​the data-driven intelligent design method is to establish a mathematical model that can efficiently calculate the service performance of structural parameters through machine learning or deep learning modeling methods, and combine it with the global optimization design method to solve it, so as to obtain the optimal structural parameters. Compared with the trial-and-error design method, the data-driven intelligent design method can greatly shorten the lengthy development cycle, reduce the high experimental costs, and often find a better global optimal solution for structural parameters, providing guidance for the application of SiC fiber-reinforced metal matrix composite tubes in the aerospace field.

[0022] The advantages and beneficial effects of the present invention are:

[0023] Aiming at the parameter design problem of SiC fiber reinforced metal matrix composite tube structure, the present invention proposes a data-driven intelligent design method for SiC fiber reinforced metal matrix composite tube structure. Compared with the existing design methods, the present invention introduces data-driven design ideas, finite element simulation analysis methods, machine learning modeling methods, global optimization algorithms and other technologies, which can efficiently and accurately design SiC fiber reinforced metal matrix composite tube structures, shorten the development cycle of composite tubular structural parts, improve the service performance of composite tubular structural parts, and enhance the application value of SiC fiber reinforced metal matrix composite tube structures in the aerospace field. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 It is a schematic diagram of the process of the present invention.

[0025] Figure 2 Schematic diagram of SiC fiber reinforced metal matrix composite tube structure. (a) is a stereoscopic diagram, (b) is a cross-sectional diagram of the center section perpendicular to the axial direction, and (c) is a partial cross-sectional diagram of the joint section parallel to the axial direction. In the figure, 1 is the inner metal matrix, 2 is the composite material layer, 3 is the outer metal matrix, 4 is SiC fiber, 5 is the center section, and 6 is the joint section.

[0026] Figure 3 Schematic diagram of the structural parameters of the SiC fiber reinforced metal matrix composite tube structure. DETAILED DESCRIPTION

[0027] In the specific implementation process, the present invention proposes a data-driven intelligent design method for SiC fiber reinforced metal matrix composite pipe structure, which includes six steps: first, the structural parameters to be designed in the pipe structure are defined as design variables, and the weight of the pipe structure and the service performance under different working conditions are divided into design objectives and design constraints according to the needs; then, an experimental design is performed, and based on the experimental design algorithm, multiple structural parameter combinations to be simulated are obtained; then, data acquisition is performed, and based on the finite element calculation method, the structural parameter combinations obtained by the experimental design are simulated to obtain the service performance of the pipe structure corresponding to each structural parameter combination; then, model training is performed, and the structural parameters are used as data features and the service performance of the corresponding pipe structure is used as data labels to train a machine learning model for predicting the service performance of the pipe structure; then, optimization and solution are performed, and the solution is performed based on the optimization algorithm. During the solution process, the model trained in the model training is used to calculate the required service performance results, so as to obtain the optimal structural parameter combination; finally, the result verification is performed, the corresponding SiC fiber reinforced metal matrix composite pipe structure is prepared, and a service performance test is performed to verify the design result.

[0028] like Figure 2 As shown in FIG. 1 , the SiC fiber reinforced metal matrix composite tube structure is a tubular structure formed by coaxially compounding an inner metal matrix 1, a composite material layer 2, and an outer metal matrix 3 in sequence. Figure 2 As shown in (a), the outer shape of the tube structure is a cylinder of equal outer diameter, and a SiC fiber reinforced metal matrix composite material layer is provided in the tube wall; wherein, the outer shape of the metal matrix tube structure is a cylinder of equal outer diameter, and the tube wall thickness ranges from 4 to 8 mm, and the material includes but is not limited to aluminum alloy, titanium alloy or titanium aluminum alloy, etc.; in the composite material layer 2, SiC fibers 4 with metal coating deposited on the surface are arranged in the metal matrix tube wall as reinforcement. The tube structure is divided into a central section 5 and a joint section 6. The joint section 6 is the part where the two ends of the composite material layer 2 are connected to the tube structure. The central section 5 is the part of the SiC fiber reinforced metal matrix composite tube structure where the SiC fibers 4 are evenly arranged, and is not the joint section 6. The cross-sectional view of the central section 5 perpendicular to the axial direction is shown in FIG. Figure 2 As shown in (b), the SiC fiber reinforced metal matrix composite material layer 2 is arranged in the metal matrix tube wall composed of the inner metal matrix 1 and the outer metal matrix 3. The thickness of the composite material layer 2 ranges from 2 to 4 mm. The composite material layer 2 is equidistant from the inner and outer walls of the tube structure. No SiC fibers 4 are exposed above or below the tube structure. The axial direction of the SiC fibers 4 is consistent with the axial direction of the tube structure. The SiC fibers 4 are arranged in layers along the radial direction of the tube structure. The number of SiC fiber layers ranges from 4 to 8 layers. The SiC fibers between the layers are densely arranged in an interlaced manner, that is, the SiC fibers in the cross section of the tube structure are approximately hexagonally densely packed. The cross section of the joint segment 6 parallel to the axial direction is as follows: Figure 2As shown in (c), the two ends of the composite material layer 2 are arranged in a step-like manner at the joint of the pipe structure, and the length of each layer of SiC fiber 4 decreases layer by layer from the inside to the outside, and the single-side decreasing range of each layer is 0.5-2.5 mm, so that the angle α between the bottom edge connection line or the top edge connection line of the composite material layer 2 and the axial edge is in the range of 5-20°. The wall thickness of the pipe structure joint section 6 is greater than the wall thickness of the center section 5, and the thickness of the inner metal matrix 1 of the pipe structure joint section 6 is greater than the thickness of the inner metal matrix 1 of the center section 5, and the range is greater than 50-100% of the thickness of the inner metal matrix 1 of the center section 5. The end of the thickened part of the inner metal matrix 1 is flush with the end of the outermost layer of SiC fiber 4 of the composite material layer 2, and the different thickness areas of the inner metal matrix 1 are connected by chamfering.

[0029] In order to make the purpose, technical solution and advantages of the present invention more clear, the present invention is further described in detail by taking a certain SiC fiber reinforced metal matrix composite material tube structure design as an example. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0030] Example

[0031] like Figure 1 As shown, a data-driven intelligent design method for SiC fiber reinforced metal matrix composite pipe structure includes the following steps:

[0032] Step 1: Problem definition. Specifically, define the design problem of SiC fiber reinforced metal matrix composite tube structure, including: 1) define the structural parameters of the tube structure to be designed and the range of the parameters. In this embodiment, the inner metal matrix thickness h is defined as in , the thickness of the outer metal substrate h out , pipe structure joint section increases wall thickness h head , thickness of composite material layer h c , the length difference of each fiber layer in the composite material layer is Δd c , the distance d between the longest fiber layer in the composite material layer and the two ends of the tube structure head Six parameters are used as design variables; 2) define the design objectives and design constraints of the pipe structure. In this embodiment, the design objective is to minimize the weight of the pipe structure G, and the maximum load F that the pipe structure can withstand under axial tension is defined. s ≮1800kN, maximum load F that the pipe structure can withstand under axial compression conditions c ≮300kN is the design constraint, and the expression of the design problem is:

[0033] minG

[0034] findh in ,h out ,h head ,h c,Δd c ,d head

[0035] ST.F s ≮1800kN

[0036] F c ≮300kN

[0037] Step 2, experimental design. Specifically, according to the defined design problem, the amount of data required for building the model is determined to be 60 groups, and the structural parameter combination of the data is determined based on the experimental design algorithm method. The experimental design algorithm includes but is not limited to full factor sampling, orthogonal experimental design, Latin hypercube sampling, optimal Latin hypercube sampling, etc. In this embodiment, the experimental design algorithm is the optimal Latin hypercube algorithm, and the use process of the selected experimental design algorithm is to write an algorithm program, input the amount of experimental data (60 groups in this embodiment) and the amount of structural parameters (6 in this embodiment), and the algorithm runs to output the experimental design plan, that is, the data characteristics of the data set required for modeling.

[0038] Step 3: Data acquisition. Specifically, based on the finite element simulation calculation method, the weight G of the corresponding pipe structure of the 60 sets of structural parameter combinations in the data set and the maximum load F that the pipe structure can withstand under axial tension conditions are calculated. s , the maximum load F that the pipe structure can withstand under axial compression conditions c The simulation calculation includes the steps of constructing the geometric model of the pipe structure, constructing the finite element simulation pre-processing model, calculating and solving the finite element simulation model, and post-processing the finite element simulation results. After the calculation is completed, the structural parameter combination [h in ,h out ,h head ,h c ,Δd c ,d head ] as data features, and service performance [G,F s ,F c ] are data labels, which constitute the data set required for modeling.

[0039] Step 4: Model training. Based on the machine learning modeling method, using the data set established in step 3, the training input is [h in ,h out ,h head ,h c ,Δd c ,d head ], the outputs are G, F s ,F cThe three machine learning models include but are not limited to support vector regression, neural network, Kriging model, gradient boosting tree, random forest, etc. In this embodiment, the machine learning modeling method is a neural network, and the process of using the selected modeling algorithm is to write an algorithm program, input a training data set, and run the algorithm to train the neural network model to obtain a machine learning model for replacing experiments and simulations.

[0040] Step 5, optimization solution. For the problem defined in step 1, the solution is performed based on a global optimization algorithm, and the optimization algorithm includes but is not limited to a genetic algorithm, a differential algorithm, a particle swarm algorithm, etc.; in this embodiment, the optimization algorithm is a particle swarm algorithm. During the solution process, the model trained in step 4 is used to calculate the values ​​of the design objectives and design constraints, and multiple rounds of iterations are performed to obtain a structural parameter combination that optimizes the design objectives and satisfies the boundary values ​​of all design constraints.

[0041] Step 6: Result verification. Prepare the SiC fiber reinforced metal matrix composite tube structure corresponding to the structural parameter combination obtained in step 5, and conduct a service performance test. If the maximum load F that the tube structure can withstand under the axial tensile condition is met, s ≮1800kN, maximum load F that the pipe structure can withstand under axial compression conditions c If the design constraint is not met, repeat steps 2 to 5 to increase the data volume and update the machine learning model until the final design result is obtained.

[0042] In this embodiment, the time and cost for obtaining the maximum load of a single set of axial tension conditions are shown in Table 1:

[0043] Table 1 Time and cost for obtaining the maximum load under axial tension condition

[0044] How to obtain Get time Acquisition Cost Experimental Testing 3-6 months 50,000-200,000 yuan Finite element simulation 1-2 hours Ignore Machine Learning Models 0.1-0.5 seconds Ignore

[0045] Obviously, finite element simulation and machine learning models can obtain the service performance of the tube structure at negligible cost during the optimization design process. Therefore, compared with the traditional trial-and-error design method, the present invention can design SiC fiber reinforced metal matrix composite tube structures with higher efficiency and lower cost, shorten the development cycle of composite tubular structural parts, improve the service performance of composite tubular structural parts, and enhance the application value of SiC fiber reinforced metal matrix composite tube structures in the aerospace field.

[0046] The above-mentioned specific implementation can be partially adjusted in different ways by those skilled in the art without departing from the principle and purpose of the present invention. The protection scope of the present invention shall be based on the claims and shall not be limited by the above-mentioned specific implementation. Each implementation scheme within its scope shall be subject to the constraints of the present invention.

Claims

1. A data-driven intelligent design method for SiC fiber reinforced metal matrix composite pipe structure, characterized in that: The steps include: Step 1 is problem definition, which defines the structural parameters to be designed in the pipe structure as design variables, and divides the weight of the pipe structure and the service performance under different working conditions into design objectives and design constraints according to requirements; Step 2 is experimental design, which obtains multiple structural parameter combinations to be simulated based on the experimental design algorithm; Step 3 is data acquisition. Based on the finite element calculation method, the structural parameter combinations obtained by the experimental design are simulated to obtain the service performance of the pipe structure corresponding to each structural parameter combination; Step 4 is model training, which uses the structural parameters as data features and the service performance of the corresponding pipe structure as data labels to train a machine learning model for predicting the service performance of the pipe structure; Step 5 is optimization solution, which is solved based on the optimization algorithm. During the solution process, the model trained in step 4 is used to calculate the required service performance results, so as to obtain the optimal combination of structural parameters; Step 6 is result verification. The SiC fiber reinforced metal matrix composite tube structure corresponding to the structural parameter combination obtained in step 5 is prepared and a service performance test is carried out. If it meets the design index requirements, the final design result is obtained; if it does not meet the design index requirements, steps 2 to 5 are repeated to increase the data volume and update the machine learning model until the final design result is obtained.

2. The method according to claim 1, characterized in that: The SiC fiber reinforced metal matrix composite material tube structure is a tubular structure formed by coaxially compounding an inner metal matrix, a composite material layer, and an outer metal matrix in sequence. The outer shape of the tube structure is a cylinder of equal outer diameter. The SiC fiber reinforced metal matrix composite material layer is arranged in the metal matrix tube wall composed of the inner metal matrix and the outer metal matrix. In the composite material layer, SiC fibers with metal coatings deposited on the surface are arranged in the metal matrix tube wall as reinforcements, and the two ends of the composite material layer and the tube structure joint are arranged in a stepped manner. The tube structure is divided into a central section and a joint section. The joint section is the part where the two ends of the composite material layer are connected to the tube structure. The central section is the non-joint section part of the SiC fiber reinforced metal matrix composite material tube structure where SiC fibers are uniformly arranged. The wall thickness of the tube structure joint section is greater than the wall thickness of the central section.

3. The method according to claim 1, characterized in that: The problem definition in step 1 refers to: defining the design problem of SiC fiber reinforced metal matrix composite tube structure, including the following steps: 1) defining the structural parameters of the tube structure to be designed and the value range of the parameters, including but not limited to the inner metal matrix thickness h in , the thickness of the outer metal substrate h out , pipe structure joint section increases wall thickness h head , thickness of composite material layer h c , the number of fiber layers in the composite material layer N c , the length difference of each fiber layer in the composite material layer is Δd c , the distance d between the longest fiber layer in the composite material layer and the two ends of the tube structure head ; 2) define the design objectives and design constraints of the pipe structure, the variables involved are: pipe structure weight G, maximum load F that the pipe structure can withstand under axial tension s , the maximum load F that the pipe structure can withstand under axial compression conditions c , these variables are divided into design objectives and design constraints according to the usage scenarios of the pipe structure, and the boundary values ​​of the design constraints are defined.

4. The method according to claim 1, characterized in that: The experimental design in step 2 refers to: according to the defined SiC fiber reinforced metal matrix composite tube structure design problem, determining the amount of data set required to build the model, and determining the structural parameter combination of the data based on the experimental design algorithm, the algorithm includes but is not limited to full factor sampling, orthogonal experimental design, Latin hypercube sampling or optimal Latin hypercube sampling.

5. The method according to claim 1, characterized in that: The data acquisition in step 3 refers to: based on the finite element simulation calculation method, calculating the service performance of the pipe structure corresponding to each structural parameter combination in the data set, specifically including but not limited to the following steps: pipe structure geometric model construction, finite element simulation pre-processing model construction, finite element simulation model calculation and solution, finite element simulation result post-processing; then, using the structural parameter combination as the data feature and the corresponding service performance as the data label, the data set required for modeling is composed.

6. The method according to claim 1, characterized in that: The model training in step 4 refers to: based on a machine learning method, using the data set established in the data acquisition step, training a machine learning model for predicting the service performance of the pipe structure, the machine learning method includes but is not limited to support vector regression, neural network, Kriging model, gradient boosting tree or random forest.

7. The method according to claim 1, characterized in that: The optimization solution in step 5 refers to: solving the problem defined in the problem definition step based on a global optimization algorithm. The optimization algorithm includes but is not limited to a genetic algorithm, a differential algorithm, or a particle swarm algorithm. During the solution process, a machine learning model is used to calculate the values ​​of the design objectives and design constraints. Multiple rounds of iterations are performed to obtain a structural parameter combination that optimizes the design objectives and satisfies all design constraint boundary values.

Images