Methods, systems, and computers for spatial distribution similarity analysis of material parameters

Abstract

This invention discloses a method, system, and computer for spatial distribution similarity analysis of material parameters, relating to the fields of materials science and engineering. By training and applying a pre-constructed neural network model with a processed set of reference parameters and a set of parameters to be analyzed, this invention obtains reference index parameters and spatial distribution similarity index parameters. Then, based on these parameters, a quantitative analysis of the similarity of the spatial distribution of material parameters is achieved. This precisely realizes the goal of quantitatively calculating the similarity of the spatial distributions of two or more sets of material parameters based on machine learning methods and neural network technology, thereby providing a reliable theoretical basis for product preparation.

Description

Technical Field

[0001] This invention relates to the fields of materials science and engineering, and in particular to a method, system, and computer for spatial distribution similarity analysis of material parameters. Background Technology

[0002] In materials science and engineering, the structure / composition and preparation / processing of materials jointly determine their properties, which in turn affect their performance in use. Quantitatively characterizing the properties of materials during processing and testing their mechanical and physical properties after processing is essential for materials research and development and product manufacturing. Traditional trial-and-error research methods require repeated experiments from prototypes to scaled-down parts to test pieces, resulting in high research costs, poor quality control, and low yield rates, thus hindering the efficiency and application of new materials and products.

[0003] Since the beginning of the 21st century, with the rapid development of semiconductor integrated circuit technology and the continuous improvement of the computing power of microcomputers, simulation research using computer simulation methods has become an important part of the development of new materials and products. Traditional experimental methods can only obtain discrete, finite, and local parameter information; that is, they can only obtain corresponding processing parameters at the locations where sensors are placed on the sample, such as temperature information at the locations of thermocouples / resistance temperature detectors (RTDs), or mechanical or physical property information at specific locations on the sample / product. However, computational simulation methods can obtain the distribution information of material parameters in the continuous space on and inside the material surface. Furthermore, recent advancements such as digital image correlation (DIC) technology can obtain the distribution information of displacement and strain on the surface of an object. These technologies and methods can provide not only material parameter information at specific locations but also spatial distribution information between parameters at different locations, greatly increasing the amount of information available in materials research and development.

[0004] However, the resulting product is not only affected by the mechanical or physical properties of a specific location; its thickness, properties, and strength are also influenced by the spatial distribution of material parameters. However, current technologies lack systems or methods for quantitatively analyzing the spatial distribution of material parameters. They can only extract characteristic values ​​from the continuous space between the material surface and its interior for analysis, or qualitatively evaluate the spatial distribution, resulting in the loss of a large amount of information. Therefore, there is an urgent need in this field for an analytical method or system that can quantitatively assess the spatial distribution of material parameters. Summary of the Invention

[0005] The purpose of this invention is to provide a method, system, and computer for analyzing the spatial distribution similarity of material parameters, which can accurately determine the similarity of the spatial distribution of two or more sets of material parameters, thereby providing a reliable theoretical basis for product preparation.

[0006] To achieve the above objectives, the present invention provides the following solution:

[0007] A method for spatial distribution similarity analysis of material parameters, comprising:

[0008] Obtain the material parameters, material parameter processing method parameters, and material parameter type parameters of each material used in the product to be manufactured; the material parameters include: material structure / composition parameters, material preparation / processing process parameters, and material performance parameters; the material parameter processing method parameters include: dimensional information of the material parameters, the value of each dimension of the material parameters, and the standardization method of the material parameters;

[0009] The material parameters are processed based on the material parameter processing method parameters and the material parameter type parameters to obtain a reference parameter group and a parameter group to be analyzed; the number of reference parameter groups is one; the number of parameter groups to be analyzed is one or more.

[0010] Construct a neural network model; the neural network model includes an encoder and a decoder;

[0011] The neural network model is trained using the aforementioned set of reference parameters to obtain a trained neural network model.

[0012] During the training of the neural network model using the reference parameter set, the loss function value of the neural network model is determined, and this loss function value is used as a reference index parameter.

[0013] The set of parameters to be analyzed is input into the trained neural network model to obtain the output value;

[0014] Determine the loss function value between the parameter set to be analyzed and the output value, and use this loss function value as a spatial distribution similarity index parameter;

[0015] Based on the reference index parameters and the spatial distribution similarity index parameters, the spatial distribution similarity among the material parameters of each material used in the product to be produced is determined, and the spatial distribution similarity result is obtained.

[0016] Optionally, the material parameters are processed based on the material parameter processing method parameters and the material parameter type parameters to obtain a reference parameter set and a parameter set to be analyzed, specifically including:

[0017] Based on the material parameter processing method parameters, the material parameters are standardized to obtain standardized material parameters;

[0018] Based on the material parameter type parameter, all the standardized material parameters used for comparison are combined into the reference parameter group;

[0019] Based on the material parameter type parameter, each group of standardized material parameters used for comparison is combined into the parameter group to be analyzed.

[0020] Optionally, the structure of the encoder is symmetrical to the structure of the decoder, or the structure of the encoder is asymmetrical to the structure of the decoder.

[0021] Optionally, the standardization methods for the material parameters include: min-max standardization, log function transformation, atan function transformation, and z-score standardization.

[0022] According to specific embodiments provided by the present invention, the present invention discloses the following technical effects:

[0023] The spatial distribution similarity analysis method for material parameters provided by this invention trains and applies a pre-constructed neural network model using a set of reference parameters and a set of parameters to be analyzed. This yields reference index parameters and spatial distribution similarity index parameters. Then, based on these parameters, a quantitative analysis of the similarity of the spatial distribution of material parameters is achieved. This method precisely realizes the goal of quantitatively calculating the similarity of the spatial distribution of two or more sets of material parameters using machine learning methods and neural network technology, thereby providing a reliable theoretical basis for product preparation.

[0024] The present invention also provides a spatial distribution similarity analysis system for material parameters, for implementing the above-described spatial distribution similarity analysis method for material parameters; the system includes:

[0025] The parameter input module is used to acquire the material parameters, material parameter processing method parameters, and material parameter type parameters of each material used in the product to be manufactured, and to process the material parameters based on the material parameter processing method parameters and the material parameter type parameters to obtain a reference parameter set and a parameter set to be analyzed; the material parameters include: material structure / composition parameters, material preparation / processing process parameters, and material performance parameters; the material parameter processing method parameters include: dimensional information of the material parameters, the value of each dimension of the material parameters, and the standardization method of the material parameters;

[0026] The machine learning technology module, connected to the parameter input module, is used to establish an artificial neural network and to train and apply the neural network model based on the reference parameter set and the parameter set to be analyzed, so as to obtain the reference index parameters and the spatial distribution similarity index parameters.

[0027] The similarity assessment module, connected to the machine learning technology module, is used to determine the spatial distribution similarity between the material parameters of each material used in the product to be produced based on the reference index parameters and the spatial distribution similarity index parameters, and to obtain the spatial distribution similarity result.

[0028] Optionally, the parameter input module includes:

[0029] The input parameter standardization processing unit is used to standardize the material parameters based on the material parameter processing method parameters to obtain standardized material parameters;

[0030] An input parameter type determination unit, connected to the input parameter standardization processing unit, is used to combine all the standardized material parameters used for comparison into the reference parameter group based on the material parameter type parameter, and to combine each group of standardized material parameters used for comparison into the parameter group to be analyzed based on the material parameter type parameter.

[0031] A reference parameter group storage unit, connected to the input parameter type determination unit, is used to store the reference parameter group;

[0032] The parameter group storage unit is connected to the input parameter type determination unit and is used to store the parameter group to be analyzed.

[0033] Optionally, the machine learning technology module includes:

[0034] The neural network training unit is connected to the reference parameter group storage unit and is used to train the neural network model using the reference parameter group to obtain a trained neural network model. It is also used to determine the loss function value of the neural network model during the training process and use this loss function value as a reference index parameter.

[0035] The neural network storage unit is connected to the reference parameter group storage unit and the neural network training unit respectively, and is used to store the structure and parameter information of the trained neural network model;

[0036] The neural network application unit is connected to both the neural network storage unit and the parameter group storage unit to be analyzed. It is used to input the parameter group to be analyzed into the trained neural network model to obtain the output value, and to determine the loss function value between the parameter group to be analyzed and the output value, and to use this loss function value as a spatial distribution similarity index parameter.

[0037] Optionally, the similarity assessment module includes:

[0038] A reference index parameter storage unit, connected to the neural network training unit, is used to store the reference index parameters;

[0039] A spatial distribution similarity index parameter storage unit, connected to the neural network application unit, is used to store the spatial distribution similarity index parameters;

[0040] The quantitative evaluation unit is connected to the reference index parameter storage unit and the spatial distribution similarity index parameter storage unit, respectively, and is used to determine the spatial distribution similarity between the material parameters of each material used in the product to be produced based on the reference index parameters and the spatial distribution similarity index parameters, so as to obtain the spatial distribution similarity result.

[0041] The present invention also provides a computer in which a spatial distribution similarity analysis system for the material parameters provided above is embedded.

[0042] Since the technical effects of the spatial distribution similarity analysis system and computer implementation of the material parameters provided by this invention are the same as those achieved by the spatial distribution similarity analysis method of the material parameters provided above, they will not be described again here. Attached Figure Description

[0043] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0044] Figure 1 A flowchart of the spatial distribution similarity analysis method for material parameters provided in Embodiment 1 of the present invention;

[0045] Figure 2 This is a schematic diagram of the structure of the neural network model provided in an embodiment of the present invention;

[0046] Figure 3 This is a schematic diagram of the structure of the spatial distribution similarity analysis system for material parameters provided in Embodiment 2 of the present invention. Detailed Implementation

[0047] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0048] The purpose of this invention is to provide a method, system, and computer for analyzing the spatial distribution similarity of material parameters. Based on machine learning methods and neural network technology, it can quantitatively calculate the similarity of the spatial distribution of two or more sets of material parameters, thereby providing a reliable theoretical basis for product preparation.

[0049] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0050] Example 1

[0051] like Figure 1 As shown, the spatial distribution similarity analysis method for material parameters provided by this invention includes:

[0052] Step 100: Obtain the material parameters, material parameter processing method parameters, and material parameter type parameters for each material used in the product to be manufactured. Material parameters include: material structure / composition parameters, material preparation / processing parameters, and material performance parameters. Material parameter processing method parameters include: dimensional information of the material parameters, the value of each dimension of the material parameter (i.e., the magnitude of each dimension of the material parameter), and the standardization method of the material parameter. The material parameter type parameter is used to determine whether the material parameter belongs to a reference parameter or a parameter to be analyzed. When the material parameter belongs to a parameter to be analyzed, the material parameter type parameter includes which group of parameters to be analyzed the material parameter belongs to. The parameter standardization methods include, but are not limited to, no standardization, min-max normalization, log function transformation, atan function transformation, and z-score normalization (zero-mean normalization). The material structure / composition parameters include, but are not limited to, the spatial distribution of the volume / mass fraction of each component, the spatial distribution of density, and the spatial distribution of the quantified geometric parameters of each component. The material preparation / processing parameters include, but are not limited to: the spatial distribution of temperature, pressure, porosity, and degree of curing of thermosetting polymers on the material surface / interior during preparation and processing. The material performance parameters include, but are not limited to, mechanical and physical properties. Mechanical properties include, but are not limited to, the spatial distribution of strength, modulus, Poisson's ratio, and toughness. Physical properties include, but are not limited to, the spatial distribution of thermal, magnetic, electrical, optical, and nuclear physical properties. Spatial distribution includes, but is not limited to, the distribution of the values ​​and directions of the corresponding physical quantities in one-dimensional, two-dimensional, three-dimensional, or higher-dimensional space.

[0053] Step 101: Process the material parameters based on the material parameter processing method parameter and the material parameter type parameter to obtain a reference parameter set and a parameter set to be analyzed. There is one set of reference parameters. There may be one or more sets of parameter sets to be analyzed.

[0054] In this embodiment, the implementation process of step 101 may be as follows: according to the material parameter processing method parameter, the obtained material parameters are processed, and the material parameters are standardized into standardized material parameters with the same dimension, the same size of each dimension and within a certain numerical range. According to the material parameter type parameter, all the standardized material parameters used for comparison in the spatial distribution similarity comparison are combined into a reference parameter group, and each group of standardized material parameters used for comparison in the spatial distribution similarity comparison is combined into a parameter group to be analyzed. There may be one or more parameter groups to be analyzed.

[0055] Step 102: Construct the neural network model. The neural network model includes an encoder and a decoder.

[0056] For example, in this embodiment, a suitable artificial neural network with an autoencoder structure or a structure similar to an autoencoder can be established based on the reference parameter set, and the established artificial neural network serves as a neural network model. The structure of the neural network model is as follows: Figure 2 As shown, it consists of two parts: an encoder and a decoder. The neural network structures of the encoder and decoder can be symmetrical or asymmetrical.

[0057] Step 103: Train the neural network model using the reference parameter set to obtain the trained neural network model.

[0058] Step 104: During the training of the neural network model using the reference parameter set, determine the loss function value of the neural network model and use this loss function value as the reference index parameter.

[0059] Based on this, the purpose of steps 103 and 104 is to use the reference parameter set as both the input and output of the artificial neural network for training the neural network. After training, the value of the loss function is used as the reference index parameter.

[0060] Step 105: Input the set of parameters to be analyzed into the trained neural network model to obtain the output value.

[0061] Step 106: Determine the loss function value between the parameter group to be analyzed and the output value, and use this loss function value as the spatial distribution similarity index parameter.

[0062] The main purpose of steps 105 and 106 is to use a trained artificial neural network to calculate and obtain spatial distribution similarity index parameters for the parameter set to be analyzed. Specifically, the parameter set to be analyzed is used as the input of the artificial neural network, the trained artificial neural network is used to calculate the output value, and the loss function between the input and output values ​​is calculated as the spatial distribution similarity index parameters.

[0063] Step 107: Determine the spatial distribution similarity among the material parameters of each material used in the product to be manufactured based on the reference index parameter and the spatial distribution similarity index parameter, and obtain the spatial distribution similarity result. For example, compare the spatial distribution similarity index parameter with the reference index parameter. If they are almost the same, the spatial distribution of the parameter group to be analyzed is similar to that of the reference parameter group. If the spatial distribution similarity index parameter is greater than the reference index parameter, there is a certain difference between the two. The larger the spatial distribution similarity index parameter, the greater the difference in spatial distribution. If there are multiple parameter groups to be analyzed, compared with the reference parameter group, the spatial distribution similarity of the parameter group to be analyzed with a larger spatial distribution similarity index parameter is less than that of the parameter group to be analyzed with a smaller spatial distribution similarity index parameter.

[0064] In actual implementation, if the spatial distribution similarity index parameter is less than the reference index parameter, it indicates that the neural network model established in step 102 may not be applicable, or that the neural network training was not completed in step 103. In this case, the implementation process of steps 102 and 103 can be adjusted according to the actual situation.

[0065] Based on the above description, after obtaining the spatial distribution similarity results, the preparation materials can be reasonably selected with reference to the requirements of the thickness, performance, strength, etc. of the product to be produced.

[0066] Example 2

[0067] This embodiment provides a spatial distribution similarity analysis system for material parameters, used to implement the spatial distribution similarity analysis method for material parameters provided in Embodiment 1 above. For example... Figure 3 As shown, the system includes: a parameter input module, a machine learning technology module, and a similarity evaluation module.

[0068] The machine learning technology module is connected to the parameter input module. The similarity evaluation module is connected to the machine learning technology module.

[0069] The parameter input module is used to acquire the material parameters, material parameter processing method parameters, and material parameter type parameters of each material used in the product to be manufactured. It then processes the material parameters based on the material parameter processing method parameters and material parameter type parameters to obtain a set of reference parameters and one or more sets of parameters to be analyzed. Material parameters include: material structure / composition parameters, material preparation / processing parameters, and material performance parameters. Material parameter processing method parameters include: dimensional information of the material parameters, the value of each dimension of the material parameters, and the standardization method of the material parameters.

[0070] The machine learning technology module is used to build artificial neural networks and to train and apply neural network models based on reference parameter sets and parameter sets to be analyzed, so as to obtain reference index parameters and spatial distribution similarity index parameters.

[0071] The similarity assessment module is used to determine the spatial distribution similarity between the material parameters of each material used in the product to be manufactured based on the reference index parameters and the spatial distribution similarity index parameters, and to obtain the spatial distribution similarity results.

[0072] Furthermore, in this embodiment, the parameter input module may include: an input parameter standardization processing unit, an input parameter type determination unit, a reference parameter group storage unit, and a parameter group storage unit to be analyzed.

[0073] The input parameter type determination unit is connected to the input parameter standardization processing unit, the reference parameter group storage unit is connected to the input parameter type determination unit, and the parameter group storage unit to be analyzed is connected to the input parameter type determination unit.

[0074] The input parameter standardization processing unit is used to standardize the material parameters based on the material parameter processing method parameters to obtain standardized material parameters. Specifically, the input parameter standardization processing unit can receive and store the input material parameters and the material parameter processing method parameters, and standardize the input material parameters according to the input material parameter processing method parameters to obtain standardized material parameters.

[0075] The input parameter type determination unit is used to combine all standardized material parameters used for comparison into a reference parameter group based on the material parameter type parameter, and to combine each group of standardized material parameters used for comparison into a parameter group to be analyzed based on the material parameter type parameter.

[0076] The reference parameter group storage unit is used to store the reference parameter group.

[0077] The parameter group storage unit is used to store the parameter group to be analyzed.

[0078] Furthermore, the machine learning technology modules employed may include: a neural network storage unit, a neural network training unit, and a neural network application unit.

[0079] The neural network training unit is connected to the reference parameter set storage unit. The neural network storage unit is connected to both the reference parameter set storage unit and the neural network training unit. The neural network application unit is connected to both the neural network storage unit and the parameter set storage unit to be analyzed.

[0080] The neural network training unit is used to train the neural network model using a set of reference parameters to obtain a trained neural network model, and is used to determine the loss function value of the neural network model during the training process, using this loss function value as a reference index parameter.

[0081] Neural network storage units are used to store the structure and parameter information of a trained neural network model.

[0082] The neural network application unit is used to input the parameter set to be analyzed into the trained neural network model to obtain the output value, and to determine the loss function value between the parameter set to be analyzed and the output value (i.e. the loss function value between the neural network input and output values), and to use this loss function value as a spatial distribution similarity index parameter.

[0083] Furthermore, the similarity assessment module includes: a reference index parameter storage unit, a spatial distribution similarity index parameter storage unit, and a quantitative assessment unit.

[0084] The reference index parameter storage unit is connected to the neural network training unit. The spatial distribution similarity index parameter storage unit is connected to the neural network application unit. The quantitative evaluation unit is connected to both the reference index parameter storage unit and the spatial distribution similarity index parameter storage unit.

[0085] The reference index parameter storage unit is used to store reference index parameters.

[0086] The spatial distribution similarity index parameter storage unit is used to store spatial distribution similarity index parameters.

[0087] The quantitative evaluation unit is used to determine the spatial distribution similarity among the material parameters of each material used in the product to be manufactured, based on reference index parameters and spatial distribution similarity index parameters, to obtain spatial distribution similarity results. For example, by comparing the spatial distribution similarity index parameters with the reference index parameters, if they are almost identical, the spatial distributions of the parameter group to be analyzed and the reference parameter group are similar. If the spatial distribution similarity index parameter is greater than the reference index parameter, there is a certain difference between the two; the larger the spatial distribution similarity index parameter, the greater the difference in spatial distribution. If there are multiple parameter groups to be analyzed, compared with the reference parameter group, the spatial distribution similarity of the parameter group to be analyzed with a larger spatial distribution similarity index parameter is less than that of the parameter group to be analyzed with a smaller spatial distribution similarity index parameter.

[0088] Example 3

[0089] This embodiment provides a computer that is equipped with the spatial distribution similarity analysis system for material parameters provided in Embodiment 2 above.

[0090] In practical applications, the computer in this embodiment has a CPU clock speed of 2.0GHz or higher, 2 cores or more, a memory capacity of 8GB or more, and a hard disk space of 60GB or more. Utilizing the inherent computing capabilities of computers, this embodiment is easy to operate and provides reliable calculation results. Through computer analysis, it is possible to effectively obtain spatial distribution information between various material parameters using experimental or simulation methods, providing a reliable mathematical tool for the research and development of new materials and products.

[0091] Example 4:

[0092] This embodiment is based on the structures in Embodiments 2 and 3 above. It illustrates the quantitative evaluation process of the similarity between the spatial distribution of temperature field and the spatial distribution of curing degree field during the curing process of carbon fiber reinforced epoxy resin laminate by using the method provided in Embodiment 1.

[0093] The basic information of the evaluated object is as follows: the carbon fiber reinforced epoxy resin laminate has dimensions of 200mm × 200mm × 4mm. The temperature and degree of cure parameters during the curing process were obtained through computer simulation. There are two laminates in total. The first laminate was cured at 120℃ / 120min, and the second laminate was cured at 160℃ / 10min + 120℃ / 80min, with a heating / cooling rate of 2℃ / min and an ambient temperature of 25℃. The material parameters of the first laminate are: the temperature field change information inside the part over time, which is stored in the form of a dot matrix cloud map, i.e., there are 3350 sets of material parameters. Each set of material parameters is a three-dimensional matrix of size 201×201×21. Each number in the matrix represents the temperature information at the corresponding location. Each set of material parameters represents the spatial distribution information of temperature at a certain moment. The time interval between the two sets of material parameters is 3s. The material parameters of the second laminate are: information on the change of the curing degree field inside the part over time, which is stored in the form of a dot matrix cloud map. There are a total of 4725 sets of material parameters. Each set of material parameters is a 401×401×41 three-dimensional matrix. Each number in the matrix represents the curing degree information at the corresponding position. Each set of material parameters represents the spatial distribution information of the curing degree at a certain moment. The time interval between the two sets of material parameters is 2 seconds.

[0094] Based on the above, the specific implementation process is as follows:

[0095] Step 1: In the parameter standardization processing unit, the material parameters of the two laminates are processed. The parameter dimension information in the parameter processing method is 3-dimensional, with the first dimension having a size of 201, the second dimension having a size of 201, and the third dimension having a size of 21. The parameter standardization method is min-max standardization, and the parameters are not standardized overall. The dimension information of each set of material parameters in the first laminate is the same as the corresponding information in the material parameter processing method, so no dimension processing is performed. The min-max standardization method is used to standardize each set of material parameters, that is, each set of temperatures is converted to the range of [0,1]. The dimension information of the material parameters in the second laminate is different from the corresponding information in the material parameter processing method. Spatial interpolation is used to interpolate each set of material parameters into a three-dimensional matrix of size 201×201×21. At the same time, the min-max standardization method is used to standardize each set of material parameters, that is, each set of curing degree is converted to the range of [0,1].

[0096] Step 2: In the input parameter type judgment unit, the standardized material parameters of the two laminates are processed. The material parameter type parameter corresponding to the first laminate is "reference", that is, the standardized material parameters of the first laminate are stored in the reference parameter group storage unit. The material parameter type parameter corresponding to the second laminate is "one group to be analyzed at each time", that is, the standardized material parameters of the second laminate are stored in the parameter group to be analyzed storage unit. Each set of standardized material parameters is a separate group, and the name of the group is the time corresponding to this set of standardized material parameters.

[0097] Step 3: Based on the parameter dimensions of the reference parameter group, establish an artificial neural network with an input-output size of 201×201×21 in the neural network storage unit, and use mean squared error as the loss function.

[0098] Step 4: Use the standardized material parameters in the reference parameter group storage unit as both input and output of the artificial neural network. Use the neural network training unit to train the artificial neural network in the neural network storage unit. After training, store the parameters of the artificial neural network in the neural network storage unit and use the function value of the loss function as the reference index parameter, which is also stored in the reference index parameter storage unit.

[0099] Step 5: Use the standardized material parameters in the storage unit of the parameter group to be analyzed as the input of the artificial neural network. Use the neural network application unit to calculate the output and loss function value of the trained artificial neural network in the neural network storage unit. Use the loss function value as the spatial distribution similarity index parameter and store it and the corresponding standardized material parameter group name in the spatial distribution similarity index parameter storage unit.

[0100] Step 6: Use the quantitative evaluation unit to compare each set of spatial distribution similarity indexes stored in the spatial distribution similarity index parameter storage unit with the reference index parameters stored in the reference index parameter storage unit.

[0101] If a certain spatial distribution similarity index parameter is almost the same as the reference index parameter, then at the corresponding time of the spatial distribution similarity index parameter, the spatial distribution of the curing degree of the second laminate is similar to the spatial distribution of the temperature of the first laminate at most times during the curing process.

[0102] If a certain spatial distribution similarity index parameter is greater than the reference index parameter, then at the corresponding time of the spatial distribution similarity index parameter, the spatial distribution of the curing degree of the second laminate is different from the spatial distribution of the temperature of the first laminate at most times during the curing process. The larger the spatial distribution similarity index parameter, the greater the difference.

[0103] Step 7: Use the quantitative evaluation unit to compare each set of spatial distribution similarity indexes stored in the spatial distribution similarity index parameter storage unit. The larger the spatial distribution similarity index parameter, the greater the difference between the curing degree spatial distribution of the corresponding second laminate and the temperature spatial distribution of the first laminate at most moments during the curing process.

[0104] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the systems disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the descriptions are relatively simple; relevant parts can be referred to the method section.

[0105] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. Furthermore, those skilled in the art will recognize that, based on the ideas of the present invention, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A method of spatial distribution similarity analysis of material parameters, characterized by, include: Obtain the material parameters, material parameter processing method parameters, and material parameter type parameters of each material used in the product to be manufactured; The material parameters include: material structure / composition parameters, material preparation / processing parameters, and material performance parameters; the material parameter processing parameters include: dimensional information of the material parameters, the value of each dimension of the material parameters, and the standardization method of the material parameters; The material parameters are processed based on the material parameter processing method parameters and the material parameter type parameters to obtain a reference parameter group and a parameter group to be analyzed; the number of reference parameter groups is one; the number of parameter groups to be analyzed is one or more. Construct a neural network model; the neural network model includes an encoder and a decoder; The neural network model is trained using the aforementioned set of reference parameters to obtain a trained neural network model. During the training of the neural network model using the reference parameter set, the loss function value of the neural network model is determined, and this loss function value is used as a reference index parameter. The set of parameters to be analyzed is input into the trained neural network model to obtain the output value; Determine the loss function value between the parameter set to be analyzed and the output value, and use this loss function value as a spatial distribution similarity index parameter; Based on the reference index parameters and the spatial distribution similarity index parameters, the spatial distribution similarity among the material parameters of each material used in the product to be produced is determined, and the spatial distribution similarity result is obtained. Specifically, the material parameters are processed based on the material parameter processing method parameters and the material parameter type parameters to obtain a reference parameter set and a parameter set to be analyzed, including: Based on the material parameter processing method parameters, the material parameters are standardized to obtain standardized material parameters; Based on the material parameter type parameter, all the standardized material parameters used for comparison are combined into the reference parameter group; Based on the material parameter type parameter, each group of standardized material parameters used for comparison is combined into the parameter group to be analyzed.

2. The method for spatial distribution similarity analysis of material parameters according to claim 1, characterized in that, The structure of the encoder is symmetrical to the structure of the decoder, or the structure of the encoder is asymmetrical to the structure of the decoder.

3. The method for spatial distribution similarity analysis of material parameters according to claim 1, characterized in that, The standardization methods for the material parameters include: min-max standardization, log function transformation, atan function transformation, and z-score standardization.

4. A spatial distribution similarity analysis system for material parameters, characterized in that, A system for implementing the spatial distribution similarity analysis method for material parameters as described in any one of claims 1-3; the system comprises: The parameter input module is used to acquire the material parameters, material parameter processing method parameters, and material parameter type parameters of each material used in the product to be manufactured, and to process the material parameters based on the material parameter processing method parameters and the material parameter type parameters to obtain a reference parameter set and a parameter set to be analyzed; the material parameters include: material structure / composition parameters, material preparation / processing process parameters, and material performance parameters; the material parameter processing method parameters include: dimensional information of the material parameters, the value of each dimension of the material parameters, and the standardization method of the material parameters; The machine learning technology module, connected to the parameter input module, is used to establish an artificial neural network and to train and apply the neural network model based on the reference parameter set and the parameter set to be analyzed, so as to obtain the reference index parameters and the spatial distribution similarity index parameters. The similarity assessment module, connected to the machine learning technology module, is used to determine the spatial distribution similarity between the material parameters of each material used in the product to be produced based on the reference index parameters and the spatial distribution similarity index parameters, and to obtain the spatial distribution similarity result.

5. The spatial distribution similarity analysis system for material parameters according to claim 4, characterized in that, The parameter input module includes: The input parameter standardization processing unit is used to standardize the material parameters based on the material parameter processing method parameters to obtain standardized material parameters; An input parameter type determination unit, connected to the input parameter standardization processing unit, is used to combine all the standardized material parameters used for comparison into the reference parameter group based on the material parameter type parameter, and to combine each group of standardized material parameters used for comparison into the parameter group to be analyzed based on the material parameter type parameter. A reference parameter group storage unit, connected to the input parameter type determination unit, is used to store the reference parameter group; The parameter group storage unit is connected to the input parameter type determination unit and is used to store the parameter group to be analyzed.

6. The spatial distribution similarity analysis system for material parameters according to claim 5, characterized in that, The machine learning technology module includes: The neural network training unit is connected to the reference parameter group storage unit and is used to train the neural network model using the reference parameter group to obtain a trained neural network model. It is also used to determine the loss function value of the neural network model during the training process and use this loss function value as a reference index parameter. The neural network storage unit is connected to the reference parameter group storage unit and the neural network training unit respectively, and is used to store the structure and parameter information of the trained neural network model; The neural network application unit is connected to both the neural network storage unit and the parameter group storage unit to be analyzed. It is used to input the parameter group to be analyzed into the trained neural network model to obtain the output value, and to determine the loss function value between the parameter group to be analyzed and the output value, and to use this loss function value as a spatial distribution similarity index parameter.

7. The spatial distribution similarity analysis system for material parameters according to claim 6, characterized in that, The similarity assessment module includes: A reference index parameter storage unit, connected to the neural network training unit, is used to store the reference index parameters; A spatial distribution similarity index parameter storage unit, connected to the neural network application unit, is used to store the spatial distribution similarity index parameters; The quantitative evaluation unit is connected to the reference index parameter storage unit and the spatial distribution similarity index parameter storage unit, respectively, and is used to determine the spatial distribution similarity between the material parameters of each material used in the product to be produced based on the reference index parameters and the spatial distribution similarity index parameters, so as to obtain the spatial distribution similarity result.

8. A computer, characterized in that, The computer is equipped with a spatial distribution similarity analysis system for material parameters as described in any one of claims 4-7.

Images