A method for evaluating and predicting subsurface damage of fiber-reinforced composites

Abstract

The present invention provides an evaluation and prediction method for subsurface damage of fiber reinforced composite materials, comprising: preparing a subsurface damage detection sample; observing the subsurface damage degree of the composite material through a microscope, and the evaluation criteria are the maximum crack depth h1, the maximum fiber pull-out depth h2, the maximum collapse pit depth h3 and the width l1; performing multiple processing tests under preset process conditions, and statistically analyzing the maximum crack depth h1, the maximum fiber pull-out depth h2, the maximum collapse pit depth h3 and the width l1 of the composite material under different process conditions, and constructing a damage database according to different damage parameters; according to the damage database, using BP neural network training, a subsurface damage prediction model of fiber reinforced composite materials under different process parameters is established. The present invention can grasp the processing damage characteristics of fiber reinforced composite materials on the basis of small batch experiments, and greatly save the material cost in the experimental process.

Description

Technical Field

[0001] The invention relates to the technical field of composite material processing, and in particular to an evaluation and prediction method for sub-surface damage of a fiber-reinforced composite material. Background Art

[0002] Fiber-reinforced ceramic matrix composites (Ceramic Matrix Composites, CMC) include structural units such as ceramic matrix, reinforcement and interface layer, that is, fiber reinforcement is introduced into the ceramic material matrix, and through the effect of appropriate weak bonding interface layer, the shortcomings of single-phase ceramic matrix materials such as high sensitivity, poor reliability and poor toughness are overcome, and the toughening and strengthening effect of reinforcement on ceramic matrix is ​​achieved. This material has the characteristics of low density, high specific strength, good thermomechanical use function, anti-ablation ability, strong thermal insulation ability, etc. It is widely used in aerospace thermal protection components to achieve thermal insulation support integration, thereby greatly reducing system weight and improving mission efficiency.

[0003] At present, the preparation of ceramic matrix composite components mostly adopts near-net-shape forming technology. In order to meet the structural and performance requirements of the components, the blank components must be processed twice to achieve the appropriate size and precision. However, fiber-reinforced ceramic matrix composites have poor machinability and obvious anisotropy, which easily leads to serious and complex machining damage.

[0004] In order to achieve the purpose of controlling processing damage, it is necessary to establish a detection and prediction method for processing damage. At present, there are technologies related to the detection and prediction of processing damage, such as a damage detection method and system for multi-layer composite materials with publication number CN110286155B, and a detection method and system for composite material layer damage with publication number CN112763452A. However, the above damage detection methods do not consider the damage evaluation method after composite material processing and the evaluation criteria under different material performance requirements.

[0005] Based on the above, it is necessary to study a composite material subsurface damage evaluation standard and assessment method. Summary of the invention

[0006] According to the technical problems raised above, a method for evaluating and predicting subsurface damage of fiber-reinforced composite materials is provided. The invention can effectively evaluate the degree of subsurface damage of fiber-reinforced composite materials after processing, and predict the degree of damage under different process conditions, saving experimental costs and providing a technical basis for the quality control of processed surfaces. The technical means adopted by the present invention are as follows:

[0007] A method for evaluating and predicting subsurface damage of a fiber reinforced composite material comprises the following steps:

[0008] Step 1: Prepare subsurface damage detection specimens;

[0009] Step 2: Observe the subsurface damage of the composite material under a microscope, and the evaluation standard is the maximum crack depth h 1 , Maximum fiber extraction depth h 2 , maximum collapse pit depth h 3 and width l 1 ;

[0010] Step 3: Carry out multiple processing tests under preset process conditions and calculate the maximum crack depth h of the composite material under different process conditions. 1 , Maximum fiber extraction depth h 2 , maximum collapse pit depth h 3 and width l 1 , construct a damage database based on different damage parameters;

[0011] Step 4: Based on the damage database, BP neural network training is used to establish a sub-surface damage prediction model for fiber-reinforced ceramic matrix composites under different process parameters.

[0012] Furthermore, the step 1 specifically includes the following steps:

[0013] Step 11, cutting the sample: cutting the processed sample into sub-surface damage samples to be observed of appropriate sizes;

[0014] Step 12, pasting the companion wafer: cut a polished silicon wafer of the same size as the sample as a companion wafer, put the sample and the companion wafer into alcohol for ultrasonic cleaning and drying, and then paste the machined surface of the sample and the polished surface of the companion wafer together;

[0015] Step 13, cross-section grinding and polishing: Fix the pasted sample on the counterweight block, and the cross-section to be observed is perpendicular to the plane where the companion piece is located and the processing direction. Grind and polish the cross-section of the sample to be observed on a grinding and polishing machine.

[0016] Furthermore, in step 12, a certain pressure should be applied to the accompanying sheet and the sample during the pasting process to ensure that the adhesive layer between the accompanying sheet and the sample is uniform and dense.

[0017] Furthermore, in step 13, the time of each process in the grinding process is long enough and a certain amount of removal is achieved to ensure that the damage introduced by the previous process is completely removed, and the surface to be observed is smooth after polishing, and the sub-surface crack information can be clearly seen.

[0018] Furthermore, in step 3, the preset process conditions include processing mode, tool linear speed, tool feed speed, tool processing depth, and tool model.

[0019] Furthermore, in step 4, the neural network model construction process includes the following steps:

[0020] Step 41, determining an appropriate number of training samples;

[0021] Step 42, dividing the training samples;

[0022] Step 43, normalizing the input layer and output layer samples;

[0023] Step 44, designing a BP network structure;

[0024] Step 45, by selecting the number of layers, hidden layer and output layer excitation function, training function, learning rate, and learning function of the BP neural network, a sub-surface damage prediction model for fiber-reinforced ceramic matrix composites under different process parameters is established.

[0025] The present invention is based on the fiber fracture mechanism and matrix crushing mechanism during the processing of fiber-reinforced composite materials, obtains the damage evaluation standard of fiber-reinforced composite materials after processing under different performance requirements, and uses a neural network to establish a damage prediction model for fiber-reinforced composite materials after processing. The damage characteristics of fiber-reinforced composite materials processing can be mastered on the basis of small batch experiments, which greatly saves the material cost in the experimental process. At the same time, the damage prediction model under different process conditions can be supplemented and improved. In the actual production process, the evaluation standard under specific process parameters is predicted by the damage prediction model, and the optimal solution that meets the preset evaluation standard is obtained by adjusting a certain process parameter, which provides a basis for optimizing the processing process parameters, is beneficial to the surface quality control of composite material processing, and provides technical support for high-quality and efficient processing of composite materials. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0027] Figure 1 The present invention is a flow chart of the composite material damage evaluation and prediction method.

[0028] Figure 2 It is a schematic diagram of a surface damage detection sample of a composite material of the present invention.

[0029] Figure 3 is the maximum crack depth h of the composite material of the present invention 1 and maximum fiber extraction depth h 2 Schematic diagram of .

[0030] Figure 4 is the maximum collapse pit depth h of the composite material of the present invention3 and width l 1 Schematic diagram of .

[0031] Figure 5 It is a schematic diagram of the BP neural network structure of the present invention. DETAILED DESCRIPTION

[0032] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0033] like Figure 1 As shown, the embodiment of the present invention discloses a method for evaluating and predicting sub-surface damage of a fiber-reinforced composite material, which comprises the following steps:

[0034] Step 1: Prepare a subsurface damage detection sample using the sample preparation method in the cross-sectional microscopic observation method;

[0035] 1): Cut the sample. Cut the processed sample into sub-surface damage samples of appropriate size to be observed. During the cutting process of the sample to be observed, the processed surface must be carefully protected to avoid damaging the original processed surface and introducing new damage.

[0036] 2): Paste the accompanying film. Figure 2 As shown, the sample and the accompanying piece are ultrasonically cleaned and dried in alcohol, and then the machined surface of the sample and the polished surface of the accompanying piece are pasted together.

[0037] 3): Cross-section grinding and polishing. Fix the pasted sample on the counterweight block. Grind and polish the section of the sample to be observed (the section to be observed is perpendicular to the plane where the companion piece is located and the processing direction) on the grinding and polishing machine.

[0038] Step 2: Observe the subsurface damage of the composite material after grinding and polishing under a microscope. The evaluation standard is the maximum crack depth h 1 , Maximum fiber extraction depth h 2 , maximum collapse pit depth h 3 and width l 1 ;

[0039] The maximum crack depth of the composite material h 1 and maximum fiber extraction depth h 2 like Figure 3 As shown, the maximum collapse pit depth h 3 and width l 1 like Figure 4 As shown in the figure, during the processing of fiber-reinforced composite materials, cracking often occurs first because the fiber-matrix interface strength is much smaller than the fiber and matrix strength. The crack propagates along the fiber-matrix interface. After the fiber-matrix interface cracks, the fiber and the interface are debonded. If the cutting force is greater than the tensile strength of the fiber, it will break and be pulled out of the matrix. The above damage forms often occur during the processing of fiber-reinforced composite materials and are prone to cause more obvious damage defects. Therefore, the above indicators are selected as evaluation indicators for damage evaluation of fiber-reinforced composite materials.

[0040] Step 3: Statistical analysis of the maximum crack depth h of the composite material under different process conditions (processing method, tool linear speed, tool feed speed, tool processing depth, tool model) 1 , Maximum fiber extraction depth h 2 , maximum collapse pit depth h 3 and width l 1 , construct a damage database based on different damage parameters;

[0041] A process database of process parameters (such as processing mode, tool linear speed, tool feed speed, tool processing depth, tool model) and the damage parameters obtained in step 2 is established to obtain the relationship between process parameters and damage parameters.

[0042] Step 4: Based on the damage database, BP neural network training is used to establish a sub-surface damage prediction model for fiber-reinforced ceramic matrix composites under different process parameters, such as Figure 5 As shown, the specific steps include:

[0043] 1): Determine the appropriate number of training samples;

[0044] 2): Divide the training samples;

[0045] 3): Normalize the input layer and output layer samples;

[0046] 4): Design of BP network structure.

[0047] 5): Through the selection of the number of layers, hidden layer and output layer excitation function, training function, learning rate, learning function, etc. of the BP neural network, a sub-surface damage prediction model of fiber-reinforced ceramic matrix composites under different process parameters is established. The process parameters and evaluation criteria obtained in the experiment are the training set and the test set. In the actual production process, the evaluation criteria under specific process parameters are predicted by the damage prediction model. The process parameters (such as processing method, tool linear speed, tool feed speed, tool processing depth, tool model) are used as the input layer. The output layer of the BP neural network model includes the maximum crack depth h 1 , Maximum fiber extraction depth h 2, maximum collapse pit depth h 3 and width l 1 . By adjusting a certain process parameter, the optimal solution that meets the preset evaluation criteria is obtained.

[0048] This method is aimed at the needs of efficient and low-damage processing of fiber-reinforced ceramic matrix composite components. It solves the current problem of lack of sub-surface damage detection and evaluation methods for fiber-reinforced ceramic matrix composites. It can effectively predict the degree of sub-surface damage under different processing conditions, has strong applicability and reliability, and provides theoretical method support for damage inhibition in fiber-reinforced ceramic matrix composite processing.

[0049] Example 1

[0050] In this example, a processing verification experiment was carried out on a fiber-reinforced ceramic-based composite material. The cutting tool was an electroplated diamond grinding wheel. The reliability of the prediction method proposed in the present invention was verified based on the side grinding test results.

[0051] The grinding test of fiber reinforced ceramic matrix composite materials has the following steps:

[0052] Step 1: Use 100#, 500#, and 1500# cup grinding wheels to grind the workpiece surface flat, ensuring that there is no obvious damage on the workpiece surface;

[0053] Step 2: Use a 6mm diameter electroplated diamond grinding wheel to grind the workpiece sideways. Set the processing parameters: grinding depth is set to 0.03mm, feed speed is set to 10mm / min, and spindle speed is set to 4000r / min.

[0054] Step 3: Replace the workpiece, change the processing parameters, and grind the workpiece.

[0055] A method for evaluating and predicting subsurface damage of a fiber reinforced composite material comprises the following steps:

[0056] Step 1: Prepare subsurface damage detection specimens using cross-sectional microscopy.

[0057] 1) Cut the sample. Use a diamond wire saw to cut the ground sample into 5mm×5mm×10mm test pieces.

[0058] Sample, of which 5mm×10mm is the surface after grinding. Carefully protect the processed surface during cutting to avoid damaging the original processed surface and introducing new damage;

[0059] 2) Paste the companion sheet. Cut a polished silicon wafer of the same size as the sample as a companion sheet, put the sample and companion sheet into alcohol for ultrasonic cleaning, dry, and then use special glue for transmission electron microscopy to paste the grinding surface and the polished surface of the companion sheet together. During the pasting process, a certain amount of pressure should be applied to the companion sheet and the sample to ensure that the glue layer between the companion sheet and the sample is uniform and fine. The companion sheet can protect the grinding surface and avoid new damage introduced by the edge effect during the grinding process. At the same time, it can also be used as a measurement reference when measuring the damage depth.

[0060] 3) Cross-section grinding and polishing. Fix the pasted sample on the round counterweight with paraffin wax. Grind the sample cross section with W20 and W5 diamond grinding discs on the grinding and polishing machine in turn. Each process has a certain amount of removal to ensure that the damage of the previous process is completely removed. Then polish the cross section with W0.5 diamond grinding paste. The surface of the polished cross section is smooth and the sub-surface crack information can be clearly seen.

[0061] Step 2: Observe the subsurface damage of the composite material under a microscope, and the evaluation standard is the maximum crack depth h 1 , Maximum fiber extraction depth h 2 , maximum collapse pit depth h 3 and width l 1 .

[0062] Step 3: Statistical analysis of the maximum crack depth h of composite materials under different process conditions 1 , Maximum fiber extraction depth h 2 , maximum collapse pit depth h 3 and width l 1 , a damage database is constructed according to different damage parameters.

[0063] Step 4: Based on the damage database, BP neural network training is used to establish a sub-surface damage prediction model for fiber-reinforced ceramic matrix composites under different process parameters.

[0064] 1): Determine the appropriate number of training samples;

[0065] 2): Divide the training samples;

[0066] 3): Normalize the input layer and output layer samples;

[0067] 4): Design of BP network structure.

[0068] 5): By selecting the number of layers, hidden layer and output layer excitation functions, training functions, learning rates, learning functions, etc. of the BP neural network, a sub-surface damage prediction model for fiber-reinforced ceramic matrix composites under different process parameters is established.

[0069] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for evaluating and predicting subsurface damage of fiber reinforced composite materials, characterized in that: The steps include: Step 1: Prepare subsurface damage detection specimens; Step 2: Observe the subsurface damage of the composite material under a microscope, and the evaluation criteria are the maximum crack depth h1, the maximum fiber pull-out depth h2, the maximum collapse pit depth h3 and width l1; Step 3: Conduct multiple processing tests under preset process conditions, calculate the maximum crack depth h1, maximum fiber pull-out depth h2, maximum collapse pit depth h3 and width l1 of the composite material under different process conditions, and build a damage database based on different damage parameters; Step 4: Based on the damage database, a BP neural network training is used to establish a subsurface damage prediction model for fiber-reinforced ceramic matrix composites under different process parameters; The step 1 specifically includes the following steps: Step 11, cutting the sample: cutting the processed sample into sub-surface damage samples to be observed of appropriate size; Step 12, pasting the companion wafer: cut a polished silicon wafer of the same size as the sample as a companion wafer, put the sample and the companion wafer into alcohol for ultrasonic cleaning and drying, and then paste the machined surface of the sample and the polished surface of the companion wafer together; Step 13, cross-section grinding and polishing: fix the pasted sample on the counterweight block, and grind and polish the cross-section to be observed on a grinding and polishing machine while making the cross-section to be observed perpendicular to the plane where the companion piece is located and the processing direction; In the step 13, the time of each process in the grinding process is long enough and a certain amount of removal is ensured to completely remove the damage introduced by the previous process, and the surface to be observed is smooth after polishing, and the sub-surface crack information can be clearly seen; The preset process conditions include processing mode, tool linear speed, tool feed speed, tool processing depth, and tool model; In step 4, the neural network model construction process includes the following steps: Step 41, determining an appropriate number of training samples; Step 42, dividing the training samples; Step 43, normalizing the input layer and output layer samples; Step 44, designing a BP network structure; Step 45, by selecting the number of layers, hidden layer and output layer excitation function, training function, learning rate, and learning function of the BP neural network, a sub-surface damage prediction model for fiber-reinforced ceramic matrix composites under different process parameters is established.

2. The method for evaluating and predicting subsurface damage of fiber reinforced composite materials according to claim 1, characterized in that: In step 12, a certain pressure should be applied to the accompanying sheet and the sample during the pasting process to ensure that the adhesive layer between the accompanying sheet and the sample is uniform and dense.

Images