Training method and device of prediction model, prediction method and device, medium and electronic equipment

Abstract

The invention provides a training method and device of a prediction model, a prediction method and device, a medium and electronic equipment. The training method comprises the following steps: acquiring first fabric physical characteristic data, first friction coefficient data, first process parameters and first optimal sewing parameters; performing standardization processing on the first fabric physical characteristic data, the first friction coefficient data and the first process parameters to obtain standardized input characteristics; performing standardization processing on the first optimal sewing parameters to obtain standardized output characteristics; and training an initial prediction model based on the standardized input features and the standardized output features to obtain a trained prediction model. According to the training method, the influence of various factors on the fabric wrinkling problem is fully considered, and the fabric wrinkling problem can be effectively solved through the optimal sewing parameters.

Description

Technical Field

[0001] The present disclosure belongs to the field of fabric sewing, and in particular relates to a training method for a prediction model, a prediction method, a device, a medium and an electronic device. Background Art

[0002] Sewing wrinkles are an unavoidable problem during the sewing process and a major factor in reducing garment quality and functionality. Current mainstream research focuses on addressing the negative impact of sewing wrinkles by focusing on single factors, such as the fabric's physical properties and sewing parameters. However, approaches that address fabric wrinkles solely based on a single factor often fail to effectively address the problem. Summary of the Invention

[0003] The purpose of the present disclosure is to provide a prediction model training method, prediction method, device, medium and electronic equipment for solving the problem of ineffective improvement of fabric wrinkling in the prior art.

[0004] To achieve the above-mentioned objectives and other related objectives, in a first aspect, the present disclosure provides a method for training a prediction model. The training method comprises: obtaining first fabric physical property data, first friction coefficient data, first process parameters, and first optimal sewing parameters; standardizing the first fabric physical property data, the first friction coefficient data, and the first process parameters to obtain standardized input features, wherein the standardized input features include standardized fabric physical property data, standardized friction coefficient data, and standardized process parameters; standardizing the first optimal sewing parameters to obtain standardized output features, wherein the standardized output features include standardized optimal sewing parameters; and training an initial prediction model based on the standardized input features and the standardized output features to obtain a trained prediction model, wherein the trained prediction model is used to predict the optimal sewing parameters of the fabric.

[0005] By training an initial prediction model based on first fabric physical property data, first friction coefficient data, first process parameters, and first optimal sewing parameters to obtain a trained prediction model, optimal sewing parameters can be predicted. The training method fully considers the impact of multiple factors on fabric wrinkling, and the optimal sewing parameters can effectively improve fabric wrinkling.

[0006] In one embodiment of the present disclosure, the standardized input features are expressed as:

[0007]

[0008] Wherein, x represents the original input feature, μ represents the mean of the original input feature, σ represents the standard deviation of the original input feature, and z represents the standardized input feature.

[0009] In one embodiment of the present disclosure, the first fabric physical property data, the first friction coefficient data, the first process parameters and the first optimal sewing parameters are experimentally measured data, the first optimal sewing parameters are the sewing parameters with the lowest degree of fabric wrinkling in the experimentally measured data, and the sewing parameters include presser foot pressure and sewing speed.

[0010] In one embodiment of the present disclosure, the experimentally measured data further includes a sample shrinkage rate, and the sample shrinkage rate has a positive correlation with the wrinkling degree of the fabric.

[0011] In one embodiment of the present disclosure, the first-order moment estimation of the training parameter gradient and the second-order moment estimation of the training parameter gradient during the initial prediction model training process are respectively expressed as:

[0012]

[0013] Among them, m t+1 represents the first-order moment estimate of the gradient of the training parameter at the next time step, β1 represents the first hyperparameter, m t represents the first-order moment estimate of the gradient of the training parameter at the current time step, represents the gradient of the training parameters at the current time step, J(θ t ) represents the loss function of the current time step with respect to the training parameters, v t+1 represents the second-order moment estimate of the gradient of the training parameter at the next time step, β2 represents the second hyperparameter, and v t Represents the second-order moment estimate of the gradient of the training parameters at the current time step.

[0014] In a second aspect, an embodiment of the present disclosure provides a prediction method, comprising: obtaining second fabric physical property data, second friction coefficient data and second process parameters; processing the second fabric physical property data, the second friction coefficient data and the second process parameters based on the trained prediction model described in any one of the first aspects to obtain second optimal sewing parameters.

[0015] In a third aspect, an embodiment of the present disclosure provides a training device for a prediction model, comprising: a first data acquisition module for acquiring first fabric physical property data, first friction coefficient data, first process parameters and first optimal sewing parameters; an input standardization module for standardizing the first fabric physical property data, the first friction coefficient data and the first process parameters to obtain standardized input features, wherein the standardized input features include standardized fabric physical property data, standardized friction coefficient data and standardized process parameters; an output standardization module for standardizing the first optimal sewing parameters to obtain standardized output features, wherein the standardized output features include standardized optimal sewing parameters; a prediction model training module for training an initial prediction model based on the standardized input features and the standardized output features to obtain a trained prediction model, wherein the trained prediction model is used to predict the optimal sewing parameters of the fabric.

[0016] In a fourth aspect, an embodiment of the present disclosure provides a prediction device, comprising: a second data acquisition module for acquiring second fabric physical property data, second friction coefficient data and second process parameters; a prediction model prediction module for processing the second fabric physical property data, the second friction coefficient data and the second process parameters using the trained prediction model described in any one of the first aspects to obtain second optimal sewing parameters.

[0017] In a third aspect, embodiments of the present disclosure further provide a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the computer program implements the training method described in any one of the first aspect and / or the prediction method described in the second aspect.

[0018] In a fourth aspect, embodiments of the present disclosure further provide an electronic device comprising: a memory storing a computer program; and a processor communicatively coupled to the memory, configured to execute any one of the training methods described in the first aspect and / or the prediction method described in the second aspect when the computer program is invoked.

[0019] By training an initial prediction model based on first fabric physical property data, first friction coefficient data, first process parameters, and first optimal sewing parameters to obtain a trained prediction model, optimal sewing parameters can be predicted. The training method fully considers the impact of multiple factors on fabric wrinkling, and the optimal sewing parameters can effectively improve fabric wrinkling. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 A structural diagram showing a schematic diagram of the hardware structure for running the training method and the prediction method according to an embodiment of the present disclosure is shown.

[0021] Figure 2 Shown is a flowchart of a method for training a prediction model according to an embodiment of the present disclosure.

[0022] Figure 3 Flowchart of the prediction method according to an embodiment of the present disclosure.

[0023] Figure 4 This is a flowchart for predicting optimal sewing parameters of fabric based on FNN in an embodiment of the present disclosure.

[0024] Figure 5 Schematic diagram of the structure of the training device of the prediction model of the embodiment of the present disclosure.

[0025] Figure 6 Schematic diagram of the structure of the prediction device of the embodiment of the present disclosure. DETAILED DESCRIPTION

[0026] The following describes the embodiments of the present disclosure through specific examples. Those skilled in the art can easily understand the other advantages and effects of the present disclosure from the content disclosed in this specification. The present disclosure can also be implemented or applied through different specific embodiments. The details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the present disclosure. It should be noted that the following embodiments and features in the embodiments can be combined with each other unless they conflict.

[0027] It should be noted that the illustrations provided in the following embodiments are only used to schematically illustrate the basic concept of the present disclosure. Therefore, the illustrations only show components related to the present disclosure and are not drawn according to the number, shape and size of components in actual implementation. In actual implementation, the type, quantity and proportion of each component can be changed at will, and the component layout type may also be more complicated.

[0028] The technical solutions in the embodiments of the present disclosure are described in detail below with reference to the accompanying drawings in the embodiments of the present disclosure.

[0029] The principles and implementation methods of the prediction model training method, prediction method, device, medium and electronic device disclosed in the present invention will be described in detail below, so that those skilled in the art can understand the prediction model training method, prediction method, device, medium and electronic device disclosed in the present invention without creative work.

[0030] The training method and prediction method provided in the embodiment of the present application can be run in a computer electronic terminal. Figure 1 For example, Figure 1The figure shows a schematic diagram of the hardware structure for running the training method and prediction method according to an embodiment of the present application. The electronic terminal 10 includes a memory 110 and a processor 120. The processor 120 can be a central processing unit 1210 or a dedicated neural network processor 1220. The neural network processor 1220 includes a neural network implementation engine 12210 and a dedicated hardware circuit 12220. The dedicated hardware circuit 12220 includes a matrix calculation unit 122210 and a vector calculation unit 122220.

[0031] Optionally, the neural network processor 12220 is a processor that performs neural network calculations using a dedicated hardware circuit 12220, which is an integrated circuit for performing neural network calculations and includes a matrix calculation unit 122210 and a vector calculation unit 122220 for hardware execution of vector-matrix multiplication.

[0032] Optionally, the neural network implementation engine 12210 is used to generate instructions for execution by the dedicated hardware circuit 12220. When the instructions are executed by the dedicated hardware circuit 12220, the dedicated hardware circuit 12220 performs operations specified by the neural network to generate a neural network output from the received neural network input.

[0033] Figure 2 : is a flow chart showing a method for training a prediction model according to an embodiment of the present disclosure. Figure 2 As shown, this embodiment provides a prediction model training method, which can be executed by a computer processor and includes:

[0034] Step S11, obtaining first fabric physical property data, first friction coefficient data, first process parameters and first optimal sewing parameters.

[0035] Optionally, the first fabric physical property data, the first friction coefficient data, the first process parameters, and the first optimal sewing parameters are experimentally measured data, the first optimal sewing parameters being the sewing parameters that result in the lowest degree of fabric wrinkling among the experimentally measured data, and the sewing parameters including presser foot pressure and sewing speed. The first optimal sewing parameters may refer to optimal sewing parameters associated with the first fabric physical property data, the first friction coefficient data, and the first process parameters, i.e., optimal sewing parameters under the conditions of the first fabric physical property data, the first friction coefficient data, and the first process parameters.

[0036] Optionally, fabric physical property data may include density, thickness, elastic modulus, flexural stiffness, etc. The friction coefficient may refer to the friction coefficient between the fabric and the presser foot, which measures the friction between the fabric and the presser foot and directly affects material handling during sewing. The friction coefficients between different types of fabrics and the presser foot vary significantly, and the friction coefficient data is obtained through experimental measurements. Process parameters may refer to sewing process parameters, specifically including parameters used in the sewing process, such as the presser foot, sewing thread, needle size, thread tension, rotation speed, and stitch length. The needle size may refer to the model of a sewing machine needle, for example, a size 9.

[0037] Alternatively, density may refer to the ratio of the mass to the volume of a fabric, and fabrics with higher density tend to have better wrinkle resistance. For example, high-density fabrics exhibit better wrinkle resistance during the sewing process. Thickness may refer to the thickness of the fabric, which affects its deformation and force distribution during the sewing process. Thicker fabrics can effectively disperse local pressure during sewing and reduce the risk of wrinkling. The elastic modulus determines the ability of the fabric to recover from deformation after being subjected to force. A higher elastic modulus can help the fabric recover after being compressed, but if the thickness is insufficient, irreversible wrinkles will still result. Bending stiffness reflects the fabric's ability to resist deformation under force. Moderate bending stiffness can balance the fabric's wrinkle resistance and deformation adaptability. Too low or too high bending stiffness will cause problems in sewing.

[0038] Optionally, the experimentally measured data also includes a sample shrinkage rate, which is positively correlated with the degree of fabric wrinkling. The positive correlation may mean that the higher the degree of fabric wrinkling, the higher the sample shrinkage rate. The sample shrinkage rate refers to the shrinkage rate of the sample during the sewing process. In experiments, the degree of fabric wrinkling can be determined by the sample shrinkage rate.

[0039] Alternatively, the shrinkage ratio can be expressed as:

[0040]

[0041] Wherein, L0 represents the length of the fabric before sewing. This length can be flexibly set according to actual conditions and is not explicitly limited in this embodiment. For example, L0 can be 35 cm. L1 represents the length after sewing. Ss represents the shrinkage rate. The sample shrinkage rate can be the average of several shrinkage rate measurements taken on the front side of the sample (test specimen). For example, the front side of the sample can be measured three times and the average of these three shrinkage rate measurements can be taken as the sample shrinkage rate. It should also be noted that a unified method must be used during the sample measurement process to ensure accuracy.

[0042] Optionally, in this embodiment, a tribological performance research experiment between the fabric and the presser foot is also carried out, and the experiment includes:

[0043] Experiments were conducted to measure the coefficient of friction of five typical fabrics (polyester, spandex-nylon blend, flannel, cotton, and artificial leather) with different presser feet (20Cr and PTFE) under varying loads and speeds, analyzing their variations and tribological mechanisms. The experiments showed that the response of the friction coefficient to load changes varied significantly among different fabric and presser foot combinations, and that tribological performance was influenced by factors such as the fabric material, surface characteristics, and friction speed. "Cr" stands for chrome steel, and "PTFE" for polytetrafluoroethylene. Furthermore, the friction coefficient of different fabrics under varying loads showed significant differences in response to friction speed.

[0044] In this embodiment, an experiment was also conducted to study the correlation between the degree of fabric wrinkling and influencing factors. The experimental content included:

[0045] Orthogonal experiments were designed to explore the effects of parameters such as presser foot pressure and sewing speed on the wrinkling degree of different fabrics, and to study the relationship between the tribological properties between the fabric and the presser foot and the influence of the physical properties of the fabric on the wrinkling degree of the fabric.

[0046] In this example, five fabrics with significant differences in surface properties, tribological properties, and elasticity were selected: polyester, spandex-nylon blended fabric, artificial leather, flannel, and cotton. Basic fabric parameters are shown in Table 1 below. The fabrics were cut radially into 40 x 5 cm rectangular strips, and the central 35 cm area was selected for sewing experiments. To minimize error, all samples were sewn by the same operator using the same equipment, with each sample sewn three times.

[0047]

[0048] Table 1

[0049] The orthogonal experimental method was used to design the experiment. The orthogonal experimental design of this scheme is shown in Table 2 below. In order to facilitate subsequent data statistics and analysis, this study will use the test number instead of the sewing process parameters in the tables listed.

[0050]

[0051] Table 2

[0052] Subjective evaluation of sewing tests was conducted according to AATCC-88B (American Association of Textile Chemists and Colorists - Method 88B, which assesses the smoothness of seams after repeated home laundering). After the sewn specimens were allowed to rest for 24 hours, three independent measurements were taken at different times, and the average value was calculated. According to AATCC-88B, 1 indicates the most severe wrinkling, while 5 indicates no wrinkling. Table 3 below provides a detailed description of the appearance of each level of seam smoothness.

[0053]

[0054] Table 3

[0055] Alternatively, the fabric wrinkling degree can be expressed as:

[0056] Sp=6-Sm

[0057] Wherein, Sp represents the wrinkle degree of the fabric, and Sm represents the flatness grade.

[0058] The average values ​​of the subjective evaluation results of the fabric sewing wrinkle degree are shown in Table 4 below. The higher the wrinkle degree value, the greater the wrinkle degree of the sample. The average values ​​of the objective test results of the fabric seam shrinkage rate are shown in Table 5 below. The higher the seam shrinkage rate, the greater the wrinkle degree of the sample.

[0059]

[0060] Table 4

[0061]

[0062]

[0063] Table 5

[0064] It can be observed that as the degree of wrinkling increases, the seam shrinkage rate also gradually increases. The trend of change in the degree of wrinkling and the seam shrinkage rate is consistent, indicating that the subjective and objective evaluation results are consistent. This shows that in the process of evaluating the degree of wrinkling and seam shrinkage of fabrics, subjective perception is consistent with experimental data, verifying the effectiveness and reliability of this method. Therefore, it can be confirmed that this evaluation method can truly reflect the seam shrinkage rate and degree of wrinkling of fabrics.

[0065] Step S12: standardize the first fabric physical property data, the first friction coefficient data, and the first process parameters to obtain standardized input features, where the standardized input features include standardized fabric physical property data, standardized friction coefficient data, and standardized process parameters.

[0066] Optionally, the standardized input feature is expressed as:

[0067]

[0068] Wherein, x represents the original input feature, μ represents the mean of the original input feature, σ represents the standard deviation of the original input feature, and z represents the standardized input feature. The original input feature includes: the first fabric physical property data, the first friction coefficient data, and the first process parameter.

[0069] Step S13 , performing standardization processing on the first optimal sewing parameters to obtain standardized output features, wherein the standardized output features include the standardized first optimal sewing parameters.

[0070] Optionally, the standardization process of the first optimal sewing parameters is similar in principle to the standardization process of the original input features, except that the original input features are replaced by the first optimal sewing parameters, which will not be described in detail in this embodiment.

[0071] Optionally, the standardization of the original input features and the first optimal sewing parameters may use a z-score (standard score) standardization method, which will not be described in detail in this embodiment.

[0072] Step S14: training the initial prediction model based on the standardized input features and the standardized output features to obtain a trained prediction model, wherein the trained prediction model is used to predict optimal sewing parameters of the fabric.

[0073] Optionally, the initial prediction model may be an initial feedforward neural network.

[0074] Optionally, the first-order moment estimation of the training parameter gradient and the second-order moment estimation of the training parameter gradient during the initial prediction model training process are respectively expressed as:

[0075]

[0076] Among them, m t+1 represents the first-order moment estimate of the gradient of the training parameter at the next time step, β1 represents the first hyperparameter, m t represents the first-order moment estimate of the gradient of the training parameter at the current time step, represents the gradient of the training parameters at the current time step, J(θ t ) represents the loss function of the current time step with respect to the training parameters, v t+1 represents the second-order moment estimate of the gradient of the training parameter at the next time step, β2 represents the second hyperparameter, and v t represents the second-order moment estimate of the gradient of the training parameter at the current time step. The first hyperparameter is used to control the attenuation of the first-order moment estimate of the gradient, and the second hyperparameter is used to control the attenuation of the second-order moment estimate of the gradient. The first and second hyperparameters can be flexibly set according to actual conditions and are not explicitly limited in this embodiment. The training parameters may include the learnable weights and biases of the initial prediction model.

[0077] Optionally, the first-order moment estimate of the training parameter gradient at the next time step after bias correction can be expressed as:

[0078]

[0079] in, It represents the first-order moment estimate of the gradient of the training parameter at the next time step after bias correction, t+1 represents the next time step, and t represents the current time step.

[0080] Optionally, the second-order moment estimate of the training parameter gradient at the next time step after bias correction can be expressed as:

[0081]

[0082] in, Represents the second-order moment estimate of the gradient of the training parameters at the next time step after bias correction.

[0083] Optionally, the training parameters at the next time step can be expressed as:

[0084]

[0085] Among them, θ t+1 represents the training parameters for the next time step, θ t represents the training parameter at the current time step, η represents the learning rate of the training parameter, ∈ represents a constant, which is introduced to avoid division by zero errors and can generally be a small constant, such as 10 -8 η can be a minimum value (e.g. 10 -8 ) to prevent the denominator from being 0.

[0086] According to the above description, the training method includes: obtaining first fabric physical property data, first friction coefficient data, first process parameters and first optimal sewing parameters; standardizing the first fabric physical property data, the first friction coefficient data and the first process parameters to obtain standardized input features, and the standardized input features include standardized first fabric physical property data, standardized first friction coefficient data and standardized first process parameters; standardizing the first optimal sewing parameters to obtain standardized output features, and the standardized output features include standardized first optimal sewing parameters; based on the standardized input features and the standardized output features, training the initial prediction model to obtain a trained prediction model, and the trained prediction model is used to predict the optimal sewing parameters of the fabric.

[0087] By training an initial prediction model based on first fabric physical property data, first friction coefficient data, first process parameters, and first optimal sewing parameters to obtain a trained prediction model, optimal sewing parameters can be predicted. The training method fully considers the impact of multiple factors on fabric wrinkling, and the optimal sewing parameters can effectively improve fabric wrinkling.

[0088] Figure 3 : is a flow chart showing the prediction method of the embodiment of the present disclosure. Figure 3 As shown, this embodiment provides a prediction method, including:

[0089] Step S21, obtaining second fabric physical property data, second friction coefficient data and second process parameters.

[0090] Optionally, the first fabric physical property data, the first friction coefficient data and the first process parameter mentioned above can be regarded as inputs for model training, and the second fabric physical property data, the second friction coefficient data and the second process parameter can be regarded as inputs for model prediction.

[0091] Step S22: processing the second fabric physical property data, the second friction coefficient data, and the second process parameters based on the trained prediction model to obtain second optimal sewing parameters.

[0092] Optionally, the first optimal sewing parameter mentioned above can be regarded as the output of model training, and the second optimal sewing parameter can be regarded as the output of model prediction.

[0093] In one embodiment of the present disclosure, the process of predicting the optimal sewing parameters of fabrics based on a feedforward neural network (FNN) is as follows: Figure 4 As shown in Figure 1, the entire prediction process includes several key steps: data collection, data preprocessing, model building, network training, and application prediction. Through the following process, the optimal sewing parameters of fabrics can be effectively predicted, the sewing process can be optimized, and production efficiency can be improved.

[0094] In this embodiment, the FNN network structure can be a feedforward neural network (FNN) with two hidden layers. The model consists of an input layer, two hidden layers, and an output layer. The specific structure is shown in Table 6 below. Each hidden layer uses a ReLU (rectified linear unit) activation function, and the output layer uses a linear activation function for regression prediction.

[0095]

[0096] Table 6

[0097] In FNN-based prediction of optimal fabric sewing parameters, constructing a feature dataset is a key step in model training. The core task of this step is to collect and organize data on the fabric's physical properties and friction coefficient to construct an input feature set that accurately reflects the fabric's performance. Constructing a feature dataset involves the following key steps:

[0098] (1) Fabric physical properties data

[0099] The physical properties of fabrics are a key factor influencing the sewing process. When constructing a feature dataset, it is necessary to collect various physical parameters related to fabrics, such as density, thickness, elastic modulus, and bending stiffness. These physical properties are typically measured experimentally and used as input features for training FNN models.

[0100] (2) Friction coefficient data

[0101] The coefficient of friction is a measure of the friction between the fabric and the presser foot, which directly affects material handling during the sewing process. When constructing a feature dataset, the coefficient of friction is a very important input feature, especially under different sewing conditions, as its changes will directly affect the sewing parameters of the fabric.

[0102] In FNN-based prediction of optimal fabric sewing parameters, data preprocessing is a key step to ensure accurate model training and efficient prediction. The preprocessing process includes standardizing input features and output targets, as well as partitioning, integrating, and normalizing the data to ensure that the model can effectively learn from the data and make accurate predictions. The specific steps are as follows:

[0103] (1) Input feature standardization

[0104] Data on fabric physical properties (such as density, thickness, elastic modulus, and flexural stiffness) and friction coefficients have different units and numerical ranges. Directly inputting this data into a neural network can lead to unstable model training and poor results. Therefore, these input features need to be normalized so that they are trained on the same scale.

[0105] The normalized formula is as follows:

[0106]

[0107] Wherein, x is the original data, i.e., the original input feature described above; μ is the mean of the data, i.e., the mean of the original input feature; σ is the standard deviation of the data, i.e., the standard deviation of the original input feature; and z is the standardized data, i.e., the standardized input feature.

[0108] (2) Target output standardization

[0109] Similarly, to ensure that the output targets (such as presser foot pressure and sewing speed) are processed at the same scale as the input features, the target data is normalized. The normalization formula is the same as that of the input features.

[0110] (3) Dataset division

[0111] The dataset needs to be divided into training, validation, and test sets to ensure that the model can effectively adjust its parameters during training and evaluate its performance on unknown data. Typically, 80% of the data is used for training and 20% for validation and testing. When dividing the dataset, random sampling is used.

[0112] (4) Data integration

[0113] After the data is normalized and partitioned, all training and validation data are combined and used as input for the FNN model training. With this data, the FNN model can learn the relationship between input features and output targets, thereby making accurate predictions.

[0114] Choosing a suitable optimization algorithm can accelerate model training, help it quickly converge to the optimal solution, and improve the generalization ability of the model, thereby enhancing the overall performance of the model. Taking all factors into consideration, the Adam algorithm is selected to calculate the estimated values ​​of the first-order moment (mean) and second-order moment (variance) of the gradient, and adaptively adjust the learning rate to achieve efficient and stable parameter optimization. The update formula of the Adam optimizer is as follows:

[0115]

[0116] The optimal fabric sewing parameter prediction based on FNN in this embodiment has the following effects:

[0117] 1. Combining tribology with the sewing process, we try to explore the underlying principles of staggered layers from a friction perspective and find ways to circumvent the staggered layer problem.

[0118] 2. Combine machine learning, algorithms and sewing tasks to improve machine intelligence and production efficiency.

[0119] 3. Explore the relationship between fabric and sewing parameters to provide certain help for the future development of unmanned sewing.

[0120] Figure 5 Schematic diagram showing the structure of the training device of the prediction model of the embodiment of the present disclosure. Figure 5 As shown, the prediction model training device 500 includes: a first data acquisition module 510, an input standardization module 520, an output standardization module 530 and a prediction model training module 540.

[0121] The first data acquisition module 510 is used to acquire first fabric physical property data, first friction coefficient data, first process parameters and first optimal sewing parameters.

[0122] The input standardization module 520 is used to standardize the first fabric physical property data, the first friction coefficient data and the first process parameters to obtain standardized input features, which include standardized fabric physical property data, standardized friction coefficient data and standardized process parameters.

[0123] The output standardization module 530 is used to perform standardization processing on the first optimal sewing parameters to obtain standardized output features, where the standardized output features include standardized optimal sewing parameters.

[0124] The prediction model training module 540 is used to train the initial prediction model based on the standardized input features and the standardized output features to obtain a trained prediction model, and the trained prediction model is used to predict the optimal sewing parameters of the fabric.

[0125] The first data acquisition module 510, the input standardization module 520, the output standardization module 530 and the prediction model training module 540 provided in this embodiment are Figure 2 Steps S11 to S14 of the training method or the detailed steps or actions in its implementation method correspond one to one and will not be repeated here.

[0126] Figure 6 Schematic diagram showing the structure of the prediction device according to the embodiment of the present disclosure. Figure 6 As shown, the prediction device 600 includes: a second data acquisition module 610 and a prediction model prediction module 620.

[0127] The second data acquisition module 610 is used to acquire second fabric physical property data, second friction coefficient data and second process parameters.

[0128] The prediction model prediction module 620 is used to process the second fabric physical property data, the second friction coefficient data and the second process parameters using the trained prediction model to obtain the second optimal sewing parameters.

[0129] The second data acquisition module 610 and the prediction model prediction module 620 provided in this embodiment are Figure 3 Steps S21 to S22 of the prediction method or detailed steps or actions in its implementation method correspond one to one and will not be repeated here.

[0130] In the several embodiments provided in the present disclosure, it should be understood that the disclosed systems, devices or methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of modules / units is only a logical function division. There may be other division methods in actual implementation. For example, multiple modules or units can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or modules or units, which can be electrical, mechanical or other forms.

[0131] Modules / units described as separate components may or may not be physically separate, and components displayed as modules / units may or may not be physical modules, that is, they may be located in one place or distributed across multiple network elements. Some or all of the modules / units may be selected based on actual needs to achieve the objectives of the embodiments of the present disclosure. For example, the functional modules / units in the various embodiments of the present disclosure may be integrated into a single processing module, or each module / unit may exist physically separately, or two or more modules / units may be integrated into a single module / unit.

[0132] Those skilled in the art should further appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the above description has generally described the composition and steps of each example according to function. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this disclosure.

[0133] The present disclosure also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the training method, prediction method, device, medium, and electronic device of the prediction model provided in the present disclosure. A person skilled in the art will understand that all or part of the steps in the method for implementing the above embodiment can be completed by instructing the processor through a program, and the program can be stored in a computer-readable storage medium, and the storage medium is a non-transitory medium, such as a random access memory, a read-only memory, a flash memory, a hard disk, a solid-state drive, a magnetic tape, a floppy disk, an optical disc, and any combination thereof. The above storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or a data center that includes one or more available media. The available medium can be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a digital video disc (DVD)), or a semiconductor medium (e.g., a solid-state drive (SSD)).

[0134] An embodiment of the present disclosure further provides an electronic device comprising a memory and a processor. The memory is used to store computer programs. In some implementations, the memory may include a computer system readable medium in the form of a volatile memory, such as RAM and / or cache memory. The electronic device may further include other removable / non-removable, volatile / non-volatile computer system storage media. The memory may include at least one program product having a set (e.g., at least one) of program modules that are configured to perform the functions of each embodiment of the present disclosure.

[0135] The processor is connected to the memory and is used to execute the computer program stored in the memory so that the electronic device can execute the training method, prediction method, device, medium and electronic device of the prediction model provided by the embodiments of the present disclosure.

[0136] In some implementations, the processor may be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc. In other implementations, the processor may also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0137] The above embodiments are merely illustrative of the principles and effects of this disclosure and are not intended to limit this disclosure. Anyone skilled in the art may modify or alter the above embodiments without departing from the spirit and scope of this disclosure. Therefore, all equivalent modifications or alterations made by one of ordinary skill in the art without departing from the spirit and technical concepts disclosed herein shall be covered by the claims of this disclosure.

Claims

1. A method for training a prediction model, characterized in that: include: Acquiring first fabric physical property data, first friction coefficient data, first process parameters, and first optimal sewing parameters; performing normalization processing on the first fabric physical property data, the first friction coefficient data, and the first process parameter to obtain a standardized input feature, wherein the standardized input feature includes the standardized first fabric physical property data, the standardized first friction coefficient data, and the standardized first process parameter; performing a standardization process on the first optimal sewing parameter to obtain a standardized output feature, wherein the standardized output feature includes the standardized first optimal sewing parameter; Based on the standardized input features and the standardized output features, the initial prediction model is trained to obtain a trained prediction model, and the trained prediction model is used to predict optimal sewing parameters of the fabric.

2. The training method according to claim 1, characterized in that The standardized input features are expressed as: Wherein, x represents the original input feature, μ represents the mean of the original input feature, σ represents the standard deviation of the original input feature, and z represents the standardized input feature.

3. The training method according to claim 1, characterized in that The first fabric physical property data, the first friction coefficient data, the first process parameters and the first optimal sewing parameters are experimentally measured data. The first optimal sewing parameters are the sewing parameters with the lowest degree of fabric wrinkling in the experimentally measured data. The sewing parameters include presser foot pressure and sewing speed.

4. The training method according to claim 3, characterized in that The experimentally measured data also includes a sample shrinkage rate, which is positively correlated with the wrinkling degree of the fabric.

5. The training method according to claim 1, wherein: The first-order moment estimation of the training parameter gradient and the second-order moment estimation of the training parameter gradient during the initial prediction model training process are respectively expressed as: Among them, m t+1 represents the first-order moment estimate of the gradient of the training parameter at the next time step, β1 represents the first hyperparameter, m t represents the first-order moment estimate of the gradient of the training parameter at the current time step, represents the gradient of the training parameters at the current time step, J(θ t ) represents the loss function of the current time step with respect to the training parameters, v t+1 represents the second-order moment estimate of the gradient of the training parameter at the next time step, β2 represents the second hyperparameter, and v t Represents the second-order moment estimate of the gradient of the training parameters at the current time step.

6. A prediction method, characterized in that: include: acquiring second fabric physical property data, second friction coefficient data, and second process parameters; The second fabric physical property data, the second friction coefficient data and the second process parameters are processed based on the trained prediction model described in any one of claims 1 to 5 to obtain second optimal sewing parameters.

7. A training device for a prediction model, characterized in that: include: A first data acquisition module is used to acquire first fabric physical property data, first friction coefficient data, first process parameters and first optimal sewing parameters; an input standardization module, configured to perform standardization processing on the first fabric physical property data, the first friction coefficient data, and the first process parameters to obtain standardized input features, wherein the standardized input features include standardized fabric physical property data, standardized friction coefficient data, and standardized process parameters; an output standardization module, configured to perform standardization processing on the first optimal sewing parameters to obtain standardized output features, wherein the standardized output features include standardized optimal sewing parameters; The prediction model training module is used to train the initial prediction model based on the standardized input features and the standardized output features to obtain a trained prediction model, and the trained prediction model is used to predict the optimal sewing parameters of the fabric.

8. A prediction device, characterized in that: include: a second data acquisition module, configured to acquire second fabric physical property data, second friction coefficient data, and second process parameters; A prediction model prediction module is used to process the second fabric physical property data, the second friction coefficient data and the second process parameters based on the trained prediction model described in any one of claims 1-5 to obtain second optimal sewing parameters.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the training method according to any one of claims 1 to 5 and / or the prediction method according to claim 6 are implemented.

10. An electronic device, characterized in that: include: a memory configured to store an executable program; The processor is configured to call the program so that the electronic device executes the training method according to any one of claims 1 to 5 and / or the prediction method according to claim 6.

Images