Spinning product process sheet generation method and device, electronic equipment and storage medium

Abstract

The invention provides a spinning product process list generation method and device, electronic equipment and a storage medium. Relates to the technical field of data processing, in particular to the technical field of artificial intelligence and deep learning. The method comprises the steps that a first time sequence set corresponding to a technological parameter set is obtained, the technological parameter set comprises a plurality of technological parameters, all the technological parameters are equipment parameters adopted by target equipment for producing a first spinning product, the first time sequence set comprises a plurality of first time sequences, and all the technological parameters are equipment parameters adopted by target equipment for producing a second spinning product; each first time sequence represents the change of the corresponding process parameter along with time; inputting the first time sequence set into a prediction model to obtain a predicted full volume rate of the first spinning product; and under the condition that the predicted full roll rate meets a preset requirement, filling each process parameter in the process parameter set into a preset template to form a target process sheet of the first spinning product.

Description

Technical Field

[0001] The present disclosure relates to the technical field of data processing, and particularly to the technical fields of artificial intelligence and deep learning. Background Art

[0002] In the industrial scenario of the spinning process, the process sheet of the spun product is a detailed document for guiding the production process. Since the spinning equipment can produce spun products of multiple batches and properties, the process sheets of each spun product are different.

[0003] Therefore, how to generate a targeted process sheet for the spun product is a problem faced currently. Summary of the Invention

[0004] The present disclosure provides a method, an apparatus, an electronic device, and a storage medium for generating a process sheet of a spun product to solve or alleviate one or more technical problems in the related art.

[0005] In a first aspect, the present disclosure provides a method for generating a process sheet of a spun product, including:

[0006] Obtaining a first time series set corresponding to a process parameter set, where the process parameter set includes multiple process parameters, each of the process parameters being an equipment parameter used by a target device to produce a first spun product, and the first time series set includes multiple first time series, each of the first time series representing the change of the corresponding process parameter over time;

[0007] Inputting the first time series set into a prediction model to obtain the predicted full winding rate of the first spun product;

[0008] When the predicted full winding rate meets a preset requirement, filling each of the process parameters in the process parameter set into a preset template to form a target process sheet of the first spun product.

[0009] In a second aspect, the present disclosure provides an apparatus for generating a process sheet of a spun product, including:

[0010] An obtaining module, configured to obtain a first time series set corresponding to a process parameter set, where the process parameter set includes multiple process parameters, each of the process parameters being an equipment parameter used by a target device to produce a first spun product, and the first time series set includes multiple first time series, each of the first time series representing the change of the corresponding process parameter over time;

[0011] A prediction module, configured to input the first time series set into a prediction model to obtain the predicted full winding rate of the first spun product;

[0012] A processing module, configured to fill each of the process parameters in the process parameter set into a preset template to form a target process sheet for the first spinning product when the predicted full roll rate meets a preset requirement.

[0013] In a third aspect, an electronic device is provided, including:

[0014] At least one processor; and

[0015] A memory communicatively connected to the at least one processor; wherein,

[0016] The memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute any method in the embodiments of the present disclosure.

[0017] In a fourth aspect, a non-transitory computer-readable storage medium storing computer instructions is provided, wherein the computer instructions are used to cause the computer to execute any method in the embodiments of the present disclosure.

[0018] In a fifth aspect, a computer program product is provided, including a computer program, and the computer program implements any method in the embodiments of the present disclosure when executed by a processor.

[0019] It should be understood that the content described in this part is not intended to identify the key or important features of the embodiments of the present disclosure, nor is it used to limit the scope of the present disclosure. Other features of the present disclosure will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In the drawings, unless otherwise specified, the same reference numerals throughout the several views denote the same or similar components or elements. These drawings are not necessarily drawn to scale. It should be understood that these drawings only depict some embodiments provided according to the present disclosure and should not be regarded as limiting the scope of the present disclosure.

[0021] Figure 1 is a schematic flowchart of a method for generating a process sheet of a spinning product according to an embodiment of the present disclosure;

[0022] Figure 2 is a schematic diagram of a process sheet according to an embodiment of the present disclosure;

[0023] Figure 3 is another schematic flowchart of a method for generating a process sheet of a spinning product according to an embodiment of the present disclosure;

[0024] Figure 4 is a schematic diagram of a first time series according to an embodiment of the present disclosure;

[0025] Figure 5 It is a schematic structural diagram of a prediction model according to an embodiment of the present disclosure;

[0026] Figure 6 It is a schematic overall flowchart of a method for generating a process sheet of a spun product according to an embodiment of the present disclosure;

[0027] Figure 7 It is another flowchart of a method for generating a process sheet of a spun product according to an embodiment of the present disclosure;

[0028] Figure 8 It is a schematic structural diagram of a device for generating a process sheet of a spun product according to an embodiment of the present disclosure;

[0029] Figure 9 It is a block diagram of an electronic device for implementing the method for generating a process sheet of a spun product according to an embodiment of the present disclosure. Detailed implementation manners

[0030] Hereinafter, the present disclosure will be described in further detail with reference to the accompanying drawings. The same reference numerals in the drawings denote elements having the same or similar functions. Although various aspects of the embodiments are shown in the drawings, the drawings do not have to be drawn to scale unless otherwise specified.

[0031] In addition, in order to better illustrate the technical solutions of the embodiments of the present disclosure, numerous specific details are given in the following detailed implementation manners. Those skilled in the art should understand that the present disclosure can also be implemented without some specific details. In some instances, methods, means, elements, and circuits well known to those skilled in the art are not described in detail so as to highlight the gist of the present disclosure.

[0032] In addition, the terms "first" and "second" are only used for descriptive purposes and cannot be construed as indicating or implying relative importance or implicitly specifying the quantity of the indicated technical features. Thus, features defined with "first" and "second" may explicitly or implicitly include one or more of such features. In the description of the present disclosure, "a plurality" means two or more unless otherwise specifically defined.

[0033] In the industrial scenario of the spinning process, the process sheet of the spun product details the records of each step of the production process.

[0034] It should be noted that the main types of spinning products involved in the solutions of the embodiments of the present disclosure may include one or more of partially oriented yarns (POY), fully drawn yarns (FDY), polyester staple fiber, etc. For example, the types of filaments may specifically include polyester partially oriented yarns, polyester fully drawn yarns, polyester drawn yarns, polyester staple fiber, etc.

[0035] In addition, the spinning products may also include slice products such as polyethylene terephthalate (PET), polyamide 6 (PA6) chips for fiber grade, etc.

[0036] A method for generating a process sheet of a spinning product is proposed in the embodiments of the present disclosure, specifically as Figure 1 shown, and can be implemented as:

[0037] S101, obtain a first time series set corresponding to a process parameter set, where the process parameter set includes a plurality of process parameters, each process parameter is an equipment parameter used by a target device to produce a first spinning product, and the first time series set includes a plurality of first time series, and each first time series represents the change of the corresponding process parameter over time.

[0038] Taking the first spinning product as POY as an example, the corresponding process parameter set includes but is not limited to the flow rate of the metering pump, the cooling air pressure, the cooling air temperature, the oiling rate of the oil pump, the rotation speed of the first hot roller, the rotation speed of the second hot roller, the rotation speed of the chuck shaft, etc.

[0039] Among them, any of the foregoing data can be used as a process parameter. Exemplarily, the flow rate of the metering pump is one process parameter, and the oiling rate of the oil pump is another process parameter.

[0040] Among them, any first time series in the first time series set is the change of the corresponding process parameter within a preset time period, where the preset time period can be set based on the actual situation, and the embodiments of the present disclosure do not limit this. Each first time series corresponds to a process parameter.

[0041] During implementation, process parameters can be collected in real time through a programmable logic controller and industrial Internet of Things technology and stored as corresponding first time series. In addition, process parameters can also be collected at preset intervals and stored as corresponding first time series. Exemplarily, the flow rate of the metering pump is collected once every 3 s, and data with a preset collection duration of 5 minutes is collected and stored as the first time series corresponding to the flow rate of the metering pump.

[0042] S102: Input the first time series set into the prediction model to obtain the predicted full bobbin rate of the first spun product.

[0043] Among them, the full bobbin rate of the spun product refers to the proportion of the number of full bobbins of the spun product that meet the quality requirements in the total number of bobbins of the spun products produced during the spinning process, and it is an important indicator for measuring the spinning production efficiency and product quality.

[0044] S103: When the predicted full bobbin rate meets the preset requirements, fill each process parameter in the process parameter set into the preset template to form the target process sheet of the first spun product.

[0045] Among them, due to the requirements of customers and the differences in spun products, the preset requirements for the full bobbin rate are also different. Exemplarily, the grades of spun products can be divided into first-class products (AA grade), first-class products (A grade), qualified products (B grade), unqualified products (C grade), etc. The full bobbin rate of first-class products can be 90% exemplarily, and the full bobbin rate of first-class products can be 80% exemplarily.

[0046] Among them, the preset template is the process sheet without filled process parameters, and the content in the process sheet includes but is not limited to information such as product information, process flow, and process parameters.

[0047] Among them, the product information is used to describe the relevant information of the spun product itself, including but not limited to information such as the process sheet number, the number of the spun product, and the product batch number of the spun product. This part of the content can be filled in by relevant personnel based on the actual situation of the first spun product.

[0048] The process flow is used to describe the processes involved in producing the spun product. This part of the content can be filled in by relevant personnel based on the actual production processes of the first spun product.

[0049] The process parameters are used to describe the equipment parameters of the target equipment used to produce the spun product. This part is automatically filled in based on the foregoing S101 - S103.

[0050] In addition, the process sheet of the first spun product can also include quality control requirements, such as the full bobbin rate required by the customer, that is, the preset requirement. The specific content included in the process sheet can be set based on the actual situation, and the embodiments of the present disclosure do not limit this.

[0051] Among them, taking the preset template of POY as an example Figure 2 As shown, the process order number, number, product batch number, and specification are product information, and spinning, winding, etc. are processes in the process flow.

[0052] The process parameters include but are not limited to spinneret diameter, spinneret diameter ratio, fineness center value, box vaporizer temperature, pipeline vaporizer temperature, metering pump speed, oil agent type, air temperature, air pressure, etc.

[0053] During implementation, after obtaining the target process order, it can be bound to the target device based on the process order number in the target process order. Specifically, it can be bound manually, that is, in the form of manual input, or the process order number in the target process order can be recognized based on Optical Character Recognition (OCR) technology, and the binding with the target device can be realized automatically.

[0054] The target process order can be used to trace the equipment parameters and process conditions during production through the target process order bound to the target device when there is an abnormality in the first spun product, and quickly locate the cause of the problem.

[0055] It should be noted that since the target device is used to produce multiple spun products, the corresponding process order bound to the target device can be located based on the batch number of the abnormal spun product, so as to trace the equipment parameters and process conditions of the abnormal spun product during production.

[0056] In the embodiments of the present disclosure, the process parameters of the target device are automatically collected, input into the prediction model to analyze the equipment operation data, and the predicted full bobbin rate is obtained. When the predicted full bobbin rate meets the preset requirements, a process order is automatically generated. Generating a process order based on this method can reduce the dependence on manual labor. The process parameters in the process order obtained based on this method meet the preset requirements of the first spun product, and the abnormality of the first spun product caused by equipment parameter problems can be reduced as much as possible, so as to further improve production efficiency and product quality.

[0057] In some embodiments, the first time series set is input into the prediction model to obtain the predicted full bobbin rate of the first spun product, as Figure 3 shown, which can be specifically implemented as:

[0058] S301, input the first time series set into the first module in the prediction model to obtain the first feature; the first module is used to capture the features of each first time series.

[0059] In some embodiments, the first module includes a shape filter and a position encoder;

[0060] Input the first time series set into the first module of the prediction model to obtain the first feature. As shown in steps A1 - A2, it can be implemented as follows:

[0061] Step A1: Use a shape filter to extract the shape subsequence corresponding to each first time series from the first time series set.

[0062] Among them, the shape subsequence is a sequence constructed by representative key data points in the first time series, which can be understood as the key sequence extracted from the first time series.

[0063] In some embodiments, in order to obtain an accurate shape subsequence, use a shape filter to extract the shape subsequence corresponding to each first time series from the first time series set. The specific implementation is as follows:

[0064] For each first time series in the first time series set, respectively execute steps B1 - B6 to obtain the shape subsequence corresponding to each first time series:

[0065] Step B1: Select a preset number of data points in the first time series to obtain an initial perception point set. Among them, each selected data point is an initial perception point.

[0066] During the operation of the device, the process parameters are usually not fixed values and usually have a small range of fluctuations. Taking the first time series of process parameter A as an example, this first time series is as Figure 4 shown. In Figure 4 , the parameters at each time point in the time period T1 - Tk of this process parameter A are {S1, S2, S3,..., Sk}, that is, (T1, S1) is a data point, and (T2, S2) is another data point.

[0067] Among them, the preset number is greater than or equal to 2. Exemplarily, the preset number can be two. Since the first and last positions usually correspond to the starting point and ending point of the first time series. The data point at the first position is the initial state of this time series. The data point at the last position is the ending state of this time series. The data points at the first and last positions can be selected as the initial perception points, which can usually effectively understand the whole process of this first time series. Exemplarily, the data points at the first position and the last position in the first time series can be selected as the initial perception points in the initial perception point set.

[0068] The preset number can be three. Randomly select three data points from the first time series as the initial perception points in the initial perception point set. Exemplarily, the data points at the first position, the middle position, and the last position in the first time series can be selected as the initial perception points in the initial perception point set.

[0069] In summary, during implementation, the first time series can be sampled according to the data point distribution to obtain a preset number of data points. The data point distribution can be determined based on the time points corresponding to each data point.

[0070] Step B2: Determine the reconstruction distances between the initial perception points in the initial perception point set and each other data point in the first time series except the initial perception points, to obtain the reconstruction distances of each other data point.

[0071] Among them, each other data point in the first time series except the initial perception points can be used as a target point respectively to calculate the reconstruction distance of the target point.

[0072] During implementation, when calculating the reconstruction distances of each target point, it can be specifically implemented as follows: when there are two initial perception points in the initial perception point set, based on the coordinates of the two initial perception points, obtain the straight line equation of the target straight line; where the target straight line is the straight line including the two initial perception points; determine the distance from the coordinates of the target point to the straight line equation of the target straight line as the reconstruction distance of the target point.

[0073] Specifically, taking the coordinates of two initial perception points as (T1, S1) and (T2, S2) as an example, the straight line equation of the target straight line, that is, the equation of y with respect to x, is shown in Equation (1):

[0074]

[0075] Taking the coordinates of the target point as (Ti, Si) as an example, the reconstruction distance of the target point is shown in Equation (2):

[0076]

[0077] Among them, based on Equation (1), B = -1, where the target point (Ti, Si) is any one of the other data points in the first time series except the initial perception points, and can also be called the target point.

[0078] In another embodiment, when there are three or more initial perception points in the initial perception point set, each two initial perception points in the initial perception point set are constructed into a group of perception point pairs to obtain multiple groups of perception point pairs; based on the average value of the reconstruction distances of the same target point relative to multiple groups of perception point pairs, determine the reconstruction distance of the target point.

[0079] In one embodiment, for each group of perception point pairs, the method for determining the reconstruction distance of the target point is similar to the calculation method when there are two initial perception points in the initial perception point set described above, and the embodiments of the present disclosure will not be elaborated herein one by one.

[0080] Exemplarily, when the initial set of perception points includes perception point 1, perception point j, and perception point k, and the target point is perception point i, the perception point pairs include perception point pair 1 as {perception point 1(T1, S1), perception point j(Tj, Sj)}, perception point pair 2 as {perception point 1(T1, S1), perception point k(Tk, Sk)}, and perception point pair 3 as {perception point j(Tj, Sj), perception point k(Tk, Sk)}. Further, the reconstruction distance between the coordinates (Ti, Si) of the target point and the straight-line equation corresponding to perception point pair 1 is R1. By calculating in sequence, the reconstruction distance Rj between the coordinates (Ti, Si) of the target point and the straight-line equation corresponding to perception point pair 2, and the reconstruction distance Rk between the coordinates (Ti, Si) of the target point and the straight-line equation corresponding to perception point pair 3 are obtained. That is, the reconstruction distance of the target point (Ti, Si) is [(R1 + Rj + Rk) / 3].

[0081] Step B3: Screen out the data points that meet the first preset condition from the reconstruction distances of other data points, and add them to the initial set of perception points to obtain a candidate set of perception points; wherein, the first preset condition includes that the reconstruction distance is not less than the first preset threshold and / or, when the reconstruction distances are sorted from large to small, the reconstruction distance ranks within the first preset quantity.

[0082] Continuing the foregoing example, when the reconstruction distance of the target point being perception point i(Ti, Si) is [(R1 + Rj + Rk) / 3] meets the first preset condition, add this target point (Ti, Si) to the initial set of perception points. That is, the initial set of perception points is updated to perception point 1, perception point i, perception point j, and perception point k, and the candidate set of perception points is {perception point 1, perception point i, perception point j, perception point k}.

[0083] In implementation, when the first preset condition is that the reconstruction distance is not less than the first preset threshold, based on the calculations in the foregoing steps B1 - B2, the reconstruction distances of other data points of this first time series can be obtained. Add the data points corresponding to the reconstruction distances not less than the first preset threshold to the initial set of perception points to obtain a candidate set of perception points.

[0084] In implementation, when the first preset condition includes that, when the reconstruction distances are sorted from large to small, the reconstruction distance ranks within the first preset quantity. Based on the calculations in the foregoing steps B1 - B2, the reconstruction distances of other data points of this first time series can be obtained. Sort the reconstruction distances of all other data points from large to small, and add the data points corresponding to the reconstruction distances ranked within the first preset quantity to the initial set of perception points to obtain a candidate set of perception points.

[0085] The first preset condition is that the reconstruction distance is not less than the first preset threshold and, when the reconstruction distances are sorted from large to small and the reconstruction distance ranking is within the first preset quantity, based on the calculations in the foregoing steps B1 - B2, the reconstruction distances of other data points of the first time series can be obtained. Sort the reconstruction distances of all other data points in descending order, and add the data points corresponding to the reconstruction distances that are not less than the first preset threshold and are ranked within the first preset quantity to the initial set of perceived points to obtain a candidate set of perceived points.

[0086] Step B4: Obtain a specified quantity of candidate perceived points from the candidate set of perceived points to obtain an initial set of shape subsequences; wherein, the initial set of shape subsequences includes at least one initial shape subsequence, and the candidate set of perceived points includes at least one candidate perceived point.

[0087] Among them, the specified quantity is the quantity of data points required in the initial shape subsequence, and the quantity of candidate perceived points in the candidate set of perceived points is not less than the specified quantity. When the quantity of candidate perceived points in any candidate set of perceived points is less than the specified quantity, discard the candidate set of perceived points.

[0088] Among them, the specified quantity is exemplarily 3. Continuing with the previous example, when the candidate set of perceived points is {perceived point 1, perceived point i, perceived point j, perceived point k}, the initial shape subsequence can be composed of any three of them. An exemplary initial set of shape subsequences can include initial shape subsequence 1 {perceived point 1, perceived point i, perceived point k}, initial shape subsequence 2 {perceived point 1, perceived point j, perceived point k}, initial shape subsequence 3 {perceived point 1, perceived point i, perceived point j}, and initial shape subsequence 4 {perceived point i, perceived point j, perceived point k}.

[0089] Step B5: Screen out the initial shape subsequences that meet the second preset condition from the information gains of the respective initial shape subsequences in the initial set of shape subsequences to obtain an intermediate set of shape subsequences; wherein

[0090] The second preset condition includes that the information gain is not less than the second preset threshold and / or when the information gains are sorted from large to small, the information gain ranking is within the second preset quantity.

[0091] In implementation, for any initial shape subsequence, calculate the information gain. Specifically, it can be implemented as: project the initial shape subsequence to obtain a first projection of the initial shape subsequence, and project the first time series to obtain a second projection of the first time series; wherein, the second projection and the first projection are in the same dimension; based on the difference between the first projection and the second projection, obtain the information gain of the initial shape subsequence.

[0092] Among them, principal component analysis (PCA) can be used to project the initial shape subsequence and / or the first time series to obtain the first projection of the initial shape subsequence and the second projection of the first time series.

[0093] Taking the projection of the initial shape subsequence to obtain the first feature as an example, it can be specifically implemented as follows: perform normalization processing on the initial shape subsequence to obtain a standard matrix; calculate the covariance matrix of the standard matrix; perform eigenvalue decomposition on the covariance matrix to obtain the eigenvectors of the covariance matrix; form a one-dimensional first projection with the eigenvectors that meet the third preset condition, where the third preset condition can be greater than the third preset threshold and is within the first third preset number in the descending order of the eigenvalues of the eigenvectors.

[0094] Specifically, performing normalization processing on the initial shape subsequence to obtain a standard matrix can be specifically implemented as follows: convert the shape subsequence into a first matrix, and then convert the first matrix into a standard matrix.

[0095] Converting the shape subsequence into a first matrix can be implemented as follows: extract subsequences with a preset defined window size of N, and use each subsequence as a row of the first matrix, that is, the number of columns of the first matrix is N, where N is a positive integer greater than 0.

[0096] Converting the first matrix into a standard matrix can be implemented as follows: calculate the mean and standard deviation of each column in the first matrix; for the mth element in the first matrix, where the mth element is any element in the first matrix, perform the following operations respectively: subtract the mean of the column corresponding to the mth element from the mth element, and then divide by the standard deviation of the column corresponding to the mth element to obtain the standard value corresponding to the mth element.

[0097] When all elements in the first matrix obtain the standard values corresponding to each element based on the foregoing operations, sort the standard values of each element according to their original positions in the first matrix to obtain the standard matrix.

[0098] Calculating the covariance matrix of the standard matrix can be obtained based on the covariance formula of the matrix, which will not be elaborated in this embodiment of the present disclosure.

[0099] In addition, projecting the first time series to obtain the second projection of the first time series is similar to the specific operations of projecting the initial shape subsequence to obtain the first feature, and will not be elaborated one by one in this embodiment of the present disclosure.

[0100] Projection can also be achieved using Linear Discriminant Analysis (LDA). Any projection method that can project the initial shape subsequence and the first time series onto the same dimension is applicable to this disclosure, and the embodiments of this disclosure do not limit this. It should be noted that the initial shape subsequence and the first time series need to be projected using the same method.

[0101] Step B6: Screen out the shape subsequences that meet the target conditions from the set of intermediate shape subsequences as the shape subsequences corresponding to the first time series.

[0102] Among them, the target condition is to screen out the sequence that is closest to the first time series as the shape subsequence.

[0103] Screening out the shape subsequences that meet the target conditions from the set of intermediate shape subsequences as the shape subsequences corresponding to the first time series can be implemented as: calculating the similarity between each intermediate shape subsequence and the first time series, and taking the intermediate shape subsequence with the maximum similarity as the final shape subsequence.

[0104] Among them, the similarity can be calculated based on cosine similarity. Any method that can obtain the similarity between the two is applicable to this disclosure, and this disclosure does not limit this.

[0105] Based on the above operations, each first time series corresponds to its own shape subsequence, and then step A2 is executed.

[0106] In the embodiments of this disclosure, relatively representative candidate sensing points in the first time series can be screened out through the reconstruction distance, the initial shape subsequence is constructed based on the candidate sensing points, its corresponding information gain is calculated, the initial shape subsequences that meet the second preset conditions are screened out as the set of intermediate shape subsequences, and they are screened, so that the finally obtained shape subsequences can effectively represent the key information of the first time series. Therefore, compared with directly using the first time series for subsequent processing, based on this method, the obtained shape subsequences can reduce the complexity of the data calculation process while retaining important information.

[0107] Step A2: Input the shape subsequences corresponding to each first time series into the position encoder to obtain the first feature.

[0108] During implementation, the shape subsequences corresponding to each first time series are converted into a form that can be processed by the position encoder and input into the position encoder, thereby obtaining the first feature.

[0109] Among them, the position encoding can provide the relative or absolute information of the positions of the shape subsequences corresponding to the first time series in the first time series set for the prediction model. By processing the shape subsequences through the position encoder, the obtained first features not only contain shape information but also incorporate time position information, enabling the prediction model to more comprehensively understand the data.

[0110] In the embodiments of the present disclosure, using a shape filter to extract shape subsequences from the first time series set can effectively capture the features of the first time series. At the same time, generating the first features through the position encoder can significantly improve the prediction model's understanding ability of the first time series, enhancing the accuracy and stability of the prediction model.

[0111] S302, input the first time series set into the second module in the prediction model to obtain second features; the second module is used to capture the overall temporal features of the first time series set.

[0112] Among them, the temporal features are used to represent the interactions and co-variations between different process parameters.

[0113] In some embodiments, the second module includes a first convolutional neural network and a second convolutional neural network, and the first convolutional neural network and / or the second convolutional neural network are composed of a one-dimensional convolutional neural network, a batch normalization neural network, and a Gaussian error linear unit network;

[0114] Inputting the first time series set into the second module in the prediction model to obtain second features, as in steps C1 - C2, can be specifically implemented as follows:

[0115] Step C1, input each first time series in the first time series set into the first convolutional neural network to obtain the time pattern of the first time series set.

[0116] Since the first time series set may exhibit patterns such as periodicity, trend, or seasonality, the time pattern is used to represent the patterns of periodicity, trend, or seasonality of the process parameters in the first time series set.

[0117] The first convolutional neural network includes a first one-dimensional convolutional neural network (Conv1D), a first batch normalization neural network (Batch Normalization, BatchNorm), and a first Gaussian error linear unit network (GELU).

[0118] During implementation, the first one-dimensional convolutional neural network can be a Conv1D filter where d c represents the number of channels, 1×d cUsed to represent the shape of the convolutional kernel. Based on this first one-dimensional convolutional neural network, time patterns in the time series can be effectively captured, such as periodicity, trend, and other regularities.

[0119] The first batch normalization neural network can enhance the generalization ability of the second model and reduce the risk of overfitting by normalizing the input of each batch.

[0120] Since the first Gaussian error linear unit network has good non-linear characteristics, it can better capture complex features on the first time series.

[0121] Step C2: Input the time pattern of the first time series set into the second convolutional neural network to obtain the second feature.

[0122] The second convolutional neural network includes a second one-dimensional convolutional neural network (Conv1D), a second batch normalization neural network (Batch Normalization, BatchNorm), and a second Gaussian error linear unit network (GELU).

[0123] The second one-dimensional convolutional neural network can be a Conv1D filter ∈ R V×1 , where V represents the number of the first time series sets within the time step, which can be understood as the number of data points in the first time series existing in the first time series set within the time step.

[0124] Exemplarily, in the case where the first time series set includes two first time series, one first time series represents the time series of the process parameter of air temperature, and the other first time series represents the time series of the process parameter of air pressure. In the case where a certain time step is Tn, the corresponding data points are two, namely (Tn, Wn) of air temperature and (Tn, Vn) of air pressure. Then the convolutional kernel will slide in the time dimension to capture the correlation between the two parameters of air temperature and air pressure. Among them, (V, 1) represents the convolutional kernel shape. It should be noted that the convolutional kernel slides only one time step in the time dimension each time, and this method can reduce redundant calculations in the time dimension to focus on the relationship between variables. Then, combined with the second batch normalization neural network and the second Gaussian error linear unit network, the second feature is obtained.

[0125] Among them, the second batch normalization neural network and the second Gaussian error linear unit network are similar in use to the aforementioned first batch normalization neural network and the first Gaussian error linear unit network, and the embodiments of the present disclosure will not elaborate on them one by one.

[0126] In the embodiments of the present disclosure, the first convolutional neural network can effectively capture the time patterns in the first time series, and the second convolutional neural network can capture the correlations between variables. By combining the batch normalization neural network and the Gaussian error linear network, the second features of the first time series extracted by the second model are more accurate.

[0127] S303. Fuse the first feature and the second feature to obtain a fused feature.

[0128] Among them, the fusion method can be to splice the first feature and the second feature, or to perform weighted summation on the first feature and the second feature to obtain the corresponding fused feature. Any method that can fuse the two is applicable to the embodiments of the present disclosure, and the present disclosure does not limit this.

[0129] S304. Input the fused feature into the classification head in the prediction model to obtain the predicted full bobbin rate of the first spinning product.

[0130] In the embodiments of the present disclosure, through the first module in the prediction module, the time series features of a single process parameter can be captured, which is the first feature. This first feature is used to understand the independent influence of each process parameter on the full bobbin rate. At the same time, through the second module in the prediction module, the time series features of the entire time series set can be captured, that is, the second feature. The second feature is used to understand the complex relationships between process parameters and their comprehensive influence on the full bobbin rate. The combination of the two can more comprehensively reflect the dynamic characteristics of the spinning process. At the same time, based on this hierarchical extraction method, the understanding ability of the model for complex process processes can be significantly improved, thereby improving the prediction accuracy.

[0131] In summary, in the embodiments of the present disclosure, the structural diagram of the prediction model is as Figure 5 shown, including a first module, a second module, and a classification head module. The first module includes a shape filter and a position encoder, where the shape filter and the position encoder are in series. The second module includes a first convolutional neural network and a second convolutional neural network. The first convolutional neural network is composed of a first one-dimensional convolutional neural network, a first batch normalization neural network, and a first Gaussian error linear network in series in sequence. The second convolutional neural network is composed of a second one-dimensional convolutional neural network, a second batch normalization neural network, and a second Gaussian error linear network in series in sequence. The first convolutional neural network and the second convolutional neural network are in series in sequence. Input the first time series set into the first module and the second module to respectively obtain the first feature output by the first module and the second feature output by the second module. Fuse the first feature and the second feature and input them into the classification head module to obtain the predicted full bobbin rate.

[0132] In some embodiments, when the preset full bobbin rate does not meet the preset requirements, prior knowledge can be combined to determine the target process sheet of the first spinning product, which can be implemented as:

[0133] Step D1, when the predicted full roll rate does not meet the preset requirements, obtain at least one process parameter corresponding to a historical spinning product of the same product type as the first spinning product from a plurality of candidate spinning products contained in the historical knowledge graph; wherein the historical knowledge graph includes the plurality of candidate spinning products and at least one process parameter corresponding to each of the candidate spinning products.

[0134] During implementation, the same product type can be understood as the first spun product and the candidate spun product are both spun products of the same grade of a spun product. For example, they can both be AA-grade products of POY.

[0135] In addition, the historical knowledge graph may also include the full roll rate of the candidate spinning products. At least one process parameter corresponding to a historical spinning product that is of the same product type as the first spinning product and has a full roll rate that meets preset requirements is obtained from the multiple candidate spinning products included in the historical knowledge graph.

[0136] Step D2, based on at least one process parameter corresponding to the historical spinning product, update the process parameter set, and return to execute the operation of obtaining the first time series set corresponding to the process parameter set until the predicted full roll rate meets the preset requirement.

[0137] In some embodiments, the process parameter set is updated based on at least one process parameter corresponding to the historical spinning product. Specifically, it can be implemented as follows: at least one process parameter corresponding to the historical spinning product is directly used as a new process parameter set.

[0138] In other embodiments, the process parameter set is updated based on at least one process parameter corresponding to the historical spinning product. Specifically, it can be implemented as follows: at least one process parameter corresponding to the historical spinning product is merged with the process parameter set (i.e., the process parameter obtained from the target device as mentioned above) to obtain a new process parameter set.

[0139] The specific fusion method can be to perform weighted summation of the same process parameters in the two to obtain a new process parameter set. Since the process parameter set includes multiple process parameters, each process parameter is weighted summed with the process parameter corresponding to the historical spinning product to obtain the optimization direction of the process parameter. For example, taking the process parameter of cooling air pressure in the process parameter set as an example, the parameter of cooling air pressure corresponding to the historical spinning product is searched, and the two are weighted summed with preset weights to obtain the final cooling air pressure parameter as the process parameter in the new process parameter set.

[0140] In the embodiments of the present disclosure, taking the fusion of at least one process parameter corresponding to a historical spinning product with a process parameter set (i.e., the process parameters obtained from the target device as described above) to obtain a new process parameter set as an example, the overall flowchart of the process sheet generation method for the spinning product proposed in the embodiments of the present disclosure is as Figure 6 shown, and it can be specifically implemented as follows:

[0141] S601, obtain a first time series set corresponding to the process parameter set.

[0142] S602, input the first time series set into a prediction model to obtain the predicted full bobbin rate of the first spinning product.

[0143] S603, determine whether the predicted full bobbin rate meets the preset requirements. If the predicted full bobbin rate does not meet the preset requirements, execute S604; if the predicted full bobbin rate meets the preset requirements, execute S607.

[0144] S604, obtain at least one process parameter corresponding to a historical spinning product of the same product type as the first spinning product from multiple candidate spinning products included in the historical knowledge graph.

[0145] S605, update the process parameter set based on at least one process parameter corresponding to the historical spinning product.

[0146] S606, extract the first time series set of the updated process parameter set; execute S602.

[0147] In some embodiments, the process parameters in the process parameter set can be fused with the process parameters of the corresponding historical spinning product based on each time point to obtain an updated first time series set.

[0148] In another embodiment, the extraction of the first time series set of the updated process parameter set can be specifically implemented as using a target prediction time series network to predict the updated process parameter set to obtain the first time series set corresponding to the updated process parameter set, where the target prediction time series network can be a long short-term memory network.

[0149] S607, fill each process parameter in the process parameter set into a preset template to form a target process sheet for the first spinning product.

[0150] The preset weight can be set based on the actual situation, and the embodiments of the present disclosure do not limit this.

[0151] Since the set of process parameters obtained in S601 is the equipment parameters directly obtained from the target equipment when the target equipment has not started production. These equipment parameters are used to verify whether the target equipment can produce the first spinning product that meets the preset requirements. Therefore, when the predicted full bobbin rate obtained based on this set of process parameters meets the preset requirements, the equipment parameters of the target equipment do not need to be modified and can directly produce the first spinning product that meets the preset requirements. When the predicted full bobbin rate does not meet the preset requirements, the set of process parameters will be updated based on historical spinning products to obtain a set of process parameters that can produce the first spinning product that meets the preset requirements. Then, a corresponding target process sheet will be generated based on the set of process parameters that can produce the first spinning product that meets the preset requirements. In this case, the equipment parameters of the target equipment need to be adjusted to the process parameters in the target process sheet so that the target equipment can produce the first spinning product that meets the preset requirements.

[0152] In the embodiments of the present disclosure, by introducing a historical knowledge graph to introduce historical successful experiences, the process of dynamically adjusting process parameters can enable the prediction model to better adapt to changes in the production process, achieve rapid iterative adjustment of process parameters, and thus reach the preset full bobbin rate requirement faster.

[0153] In some embodiments, after the target equipment is bound to the target process sheet and during the process of producing the first spinning product based on the target process sheet, the target process sheet of the target equipment is updated based on the following method, as Figure 7 shown, and specifically can be implemented as:

[0154] S701, during the process of producing the first spinning product, obtain the physical inspection result of the product sample in the first spinning product that the target equipment is producing.

[0155] Among them, a corresponding physical inspection task can be created based on the specific product information of the first spinning product to obtain the corresponding physical inspection result.

[0156] Exemplarily, when the first spinning product is POY, the physical inspection task can include at least one of the linear density deviation rate, linear density coefficient of variation, breaking strength, breaking strength coefficient of variation, elongation at break, elongation at break coefficient of variation, etc.

[0157] Among them, the linear density deviation rate: the degree of deviation between the linear density of the product sample and the linear density of the theoretical first spinning product. The theoretical first spinning product is the expected product of the first spinning product.

[0158] The linear density coefficient of variation: the degree of uniformity of the linear density of the product sample.

[0159] The breaking strength: the maximum tensile or compressive force that the product sample can withstand before breaking under the action of force.

[0160] Coefficient of variation of breaking strength: The degree of uniformity of the breaking strength of the product sample.

[0161] Elongation at break: The ratio of the deformation amount before break to the initial length during the stretching process of the product sample.

[0162] Coefficient of variation of elongation at break: The degree of uniformity of the elongation at break of the product sample.

[0163] S702, when the physical inspection result meets the physical inspection requirements, obtain the target process parameters adopted by the target device to produce the product sample.

[0164] Among them, the physical inspection requirements can be set based on the actual situation. Exemplarily, it can be required that the coefficient of variation of linear density of AA grade of POY is less than 0.8.

[0165] In the case where the physical inspection task includes multiple subtasks (exemplary subtasks can be any one of the linear density deviation rate, coefficient of variation of linear density, breaking strength, coefficient of variation of breaking strength, elongation at break, coefficient of variation of elongation at break), the physical inspection result corresponding to the physical inspection task includes multiple, that is, each of the aforementioned subtasks corresponds to a physical inspection result. When the physical inspection results of all subtasks meet the corresponding physical inspection requirements, obtain the target process parameters adopted by the target device to produce the production sample.

[0166] S703, update the target process sheet based on the target process parameters.

[0167] Fill the target process parameters into the corresponding content in the target process sheet. When the update of the target process sheet is completed, update the historical knowledge graph based on the target process sheet, which can provide accurate prior knowledge for subsequent production.

[0168] In the embodiments of the present disclosure, by obtaining the physical inspection results of the product sample and updating the target process sheet when the physical inspection results meet the requirements, the combination of theory and practice is realized in this way, making the generated target process sheet more accurate. Based on the target process sheet, the production of the first spinning product can be realized, which can ensure that the process parameters during the production process are always in the best state, thereby improving the stability and consistency of product quality.

[0169] Based on the same technical concept, in the embodiments of the present disclosure, a device 800 for generating a process sheet of a spinning product is proposed, as Figure 8 shown, including:

[0170] An acquisition module 801, configured to acquire a first time series set corresponding to a process parameter set, where the process parameter set includes a plurality of process parameters, each of the process parameters being an equipment parameter used by a target device to produce a first spinning product, and the first time series set includes a plurality of first time series, each of the first time series representing the change of the corresponding process parameter over time;

[0171] A prediction module 802, configured to input the first time series set into a prediction model to obtain a predicted full winding rate of the first spinning product;

[0172] A processing module 803, configured to fill each of the process parameters in the process parameter set into a preset template to form a target process sheet of the first spinning product when the predicted full winding rate meets a preset requirement.

[0173] In some embodiments, the prediction module includes:

[0174] A first processing unit, configured to input the first time series set into a first module in the prediction model to obtain a first feature; the first module is used to capture the features of each first time series;

[0175] A second processing unit, configured to input the first time series set into a second module in the prediction model to obtain a second feature; the second module is used to capture the temporal features of the whole first time series set;

[0176] A fusion unit, configured to fuse the first feature and the second feature to obtain a fusion feature;

[0177] A prediction unit, configured to input the fusion feature into a classification head module in the prediction model to obtain the predicted full winding rate of the first spinning product.

[0178] In some embodiments, the first module includes a shape filter and a position encoder;

[0179] The first processing unit includes:

[0180] An extraction subunit, configured to extract shape subsequences corresponding to each of the first time series from the first time series set by using the shape filter;

[0181] An input subunit, configured to input the shape subsequences corresponding to each of the first time series into the position encoder to obtain the first feature.

[0182] In some embodiments, the extraction subunit is specifically configured to:

[0183] For each first time series in the first time series set, perform the following operations respectively:

[0184] Select a preset number of data points in the first time series to construct multiple initial perception point sets;

[0185] Determine the reconstruction distances between the initial perception points in the initial perception point set and each of the other data points in the first time series except the initial perception points, to obtain the reconstruction distances of each of the other data points;

[0186] Screen out the data points that meet the first preset condition from the reconstruction distances of each of the other data points, and add them to the initial perception point set as the candidate perception point set; wherein, the first preset condition includes that the reconstruction distance is not less than the first preset threshold and / or, when the reconstruction distances are sorted from large to small, the reconstruction distance ranks within the first preset number;

[0187] Obtain a specified number of candidate perception points from the candidate perception point set to construct an initial shape subsequence set; wherein, the initial shape subsequence set includes at least one initial shape subsequence, and the candidate perception point set includes multiple candidate perception points;

[0188] Screen out the initial shape subsequences that meet the second preset condition from the information gains of each initial shape subsequence in the initial shape subsequence set to obtain an intermediate shape subsequence set; wherein the second preset condition includes that the information gain is not less than the second preset threshold and / or, when the information gains are sorted from large to small, the information gain ranks within the second preset number;

[0189] Screen out the shape subsequences that meet the target condition from the intermediate shape subsequence set as the shape subsequence corresponding to the first time series.

[0190] In some embodiments, the second module includes a first convolutional neural network and a second convolutional neural network, and the first convolutional neural network and / or the second convolutional neural network is composed of a one-dimensional convolutional neural network, a batch normalization neural network, and a Gaussian error linear network;

[0191] The second processing unit includes:

[0192] A first acquisition subunit, configured to input each first time series in the first time series set into the first convolutional neural network to obtain the time pattern of the first time series set;

[0193] A second acquisition subunit, configured to input the time pattern of the first time series set into the second convolutional neural network to obtain the second feature.

[0194] In some embodiments, it further includes a first update module, configured to:

[0195] When the predicted full bobbin rate does not meet the preset requirements, obtain at least one process parameter corresponding to a historical spinning product of the same product type as the first spinning product from multiple candidate spinning products included in the historical knowledge graph, where the historical knowledge graph includes the multiple candidate spinning products and at least one process parameter corresponding to each of the candidate spinning products;

[0196] Based on the at least one process parameter corresponding to the historical spinning product, update the process parameter set, and return to execute the operation of obtaining the first time series set corresponding to the process parameter set until the predicted full bobbin rate meets the preset requirements.

[0197] In some embodiments, it further includes a second update module, configured to: after the target device binds the target process sheet and during the process of producing the first spinning product based on the target process sheet, update the target process sheet of the target device based on the following method:

[0198] During the process of generating the first spinning product, obtain the physical inspection result of the product sample in the first spinning product that the target device is producing;

[0199] When the physical inspection result meets the physical inspection requirements, obtain the target process parameter used by the target device to produce the product sample;

[0200] Update the target process sheet based on the target process parameter.

[0201] For the specific functions and example descriptions of each module, sub-module, and unit of the device according to the embodiments of the present disclosure, reference may be made to the relevant descriptions of the corresponding steps in the above method embodiments, which will not be elaborated here.

[0202] In the technical solution of the present disclosure, the acquisition, storage, and application of the user's personal information involved all comply with the provisions of relevant laws and regulations and do not violate public order and good customs.

[0203] Figure 9 It is a structural block diagram of an electronic device according to an embodiment of the present disclosure. As Figure 9 shown, the electronic device includes: a memory 910 and a processor 920. The memory 910 stores a computer program that can run on the processor 920. The number of the memory 910 and the processor 920 can be one or more. The memory 910 can store one or more computer programs. When the one or more computer programs are executed by the electronic device, the electronic device executes the method provided in the above method embodiment. The electronic device may further include: a communication interface 930, configured to communicate with external devices and perform data interaction and transmission.

[0204] If the memory 910, the processor 920, and the communication interface 930 are implemented independently, the memory 910, the processor 920, and the communication interface 930 can be interconnected via a bus and communicate with each other. This bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, an Extended Industry Standard Architecture (EISA) bus, or the like. This bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 9 only a thick line is used to represent it in the figure, but it does not mean that there is only one bus or one type of bus.

[0205] Optionally, in a specific implementation, if the memory 910, the processor 920, and the communication interface 930 are integrated on a single chip, the memory 910, the processor 920, and the communication interface 930 can communicate with each other through an internal interface.

[0206] It should be understood that the above-mentioned processor can be a Central Processing Unit (CPU), or other general-purpose processors, Digital Signal Processors (DSPs), Application Specific Integrated Circuits (ASICs), Field Programmable Gate Arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor, etc. It is worth noting that the processor can be a processor that supports the Advanced RISC Machines (ARM) architecture.

[0207] Further, optionally, the above-mentioned memory may include a read-only memory and a random access memory, and may also include a non-volatile random access memory. The memory may be a volatile memory or a non-volatile memory, or may include both volatile and non-volatile memories. Among them, the non-volatile memory may include a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), or a flash memory. The volatile memory may include a random access memory (RAM), which is used as an external cache. By way of example but not limitation, many forms of RAM are available. For example, static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchlink dynamic random access memory (SLDRAM), and direct rambus random access memory (DR RAM).

[0208] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer instructions are loaded and executed on a computer, the processes or functions described in the embodiments of the present disclosure are generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center by wire (e.g., coaxial cable, optical fiber, Digital Subscriber Line (DSL)) or wirelessly (e.g., infrared, Bluetooth, microwave, etc.). The computer-readable storage medium can be any available medium that can be accessed by a computer, or a data storage device such as a server or data center that includes one or more integrated available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., Digital Versatile Disc (DVD)), or a semiconductor medium (e.g., Solid State Disk (SSD)), etc. It should be noted that the computer-readable storage medium mentioned in the present disclosure can be a non-volatile storage medium, in other words, a non-transitory storage medium.

[0209] Those of ordinary skill in the art can understand that all or part of the steps to implement the above embodiments can be completed by hardware, or can be completed by a program instructing relevant hardware. The program can be stored in a computer-readable storage medium, and the storage medium mentioned above can be a read-only memory, a magnetic disk, an optical disc, or the like.

[0210] In the description of the embodiments of the present disclosure, the descriptions referring to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples", etc. mean that the specific features, structures, materials, or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present disclosure. Moreover, the specific features, structures, materials, or characteristics described can be combined in a suitable manner in any one or more embodiments or examples. In addition, without contradiction, those skilled in the art can combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples.

[0211] In the description of the embodiments of the present disclosure, unless otherwise specified, " / " means "or". For example, A / B may mean A or B. The "and / or" herein is merely a description of the relationship between related objects, indicating that there can be three relationships. For example, A and / or B may mean: A exists alone, A and B exist simultaneously, and B exists alone.

[0212] In the description of the embodiments of the present disclosure, the terms "first" and "second" are only used for descriptive purposes and cannot be construed as indicating or implying relative importance or implicitly specifying the quantity of the indicated technical features. Thus, the features defined with "first" and "second" may explicitly or implicitly include one or more of such features. In the description of the embodiments of the present disclosure, unless otherwise specified, "a plurality of" means two or more.

[0213] The above are only exemplary embodiments of the present disclosure and are not intended to limit the present disclosure. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present disclosure shall be included within the protection scope of the present disclosure.

Claims

1. A method for generating a process sheet for a spinning product, comprising: Acquire a first time series set corresponding to a process parameter set, wherein the process parameter set includes a plurality of process parameters, each of which is an equipment parameter used by a target device to produce a first spun product, and the first time series set includes a plurality of first time series, each of which represents a change of a corresponding process parameter over time; Inputting the first time series set into a prediction model to obtain a predicted full roll rate of the first spun product; When the predicted full roll rate meets the preset requirements, each process parameter in the process parameter set is filled into a preset template to form a target process sheet for the first spun product.

2. The method according to claim 1, wherein: The step of inputting the first time series set into a prediction model to obtain a predicted full roll rate of the first spun product includes: Inputting the first time series set into a first module in the prediction model to obtain a first feature; the first module is used to capture the features of each first time series; Inputting the first time series set into the second module in the prediction model to obtain a second feature; the second module is used to capture the overall time series feature of the first time series set; Fusing the first feature and the second feature to obtain a fused feature; The fusion feature is input into the classification head module in the prediction model to obtain the predicted full roll rate of the first spun product.

3. The method according to claim 2, wherein: The first module includes a shape filter and a position encoder; The step of inputting the first time series set into the first module of the prediction model to obtain the first feature includes: extracting shape subsequences corresponding to each of the first time series from the first time series set using the shape filter; The shape subsequences respectively corresponding to the first time series are input into the position encoder to obtain the first feature.

4. The method according to claim 3, wherein: The extracting the shape subsequences corresponding to the first time series respectively from the first time series set by using the shape filter includes: For each first time series in the first time series set, the following operations are performed respectively: Selecting a preset number of data points in the first time series to obtain an initial perception point set; Determine a reconstruction distance between an initial perception point in the initial perception point set and each other data point in the first time series except the initial perception point, and obtain a reconstruction distance of each other data point; Data points that meet a first preset condition are selected from the reconstruction distances of the other data points, and are added to the initial perception point set to obtain a candidate perception point set; wherein the first preset condition includes that the reconstruction distance is not less than a first preset threshold and / or when the reconstruction distances are sorted from large to small, the reconstruction distance is ranked within a first preset number; Acquire a specified number of candidate perception points from the candidate perception point set to obtain an initial shape subsequence set; wherein the initial shape subsequence set includes at least one initial shape subsequence, and the candidate perception point set includes a plurality of candidate perception points; The initial shape subsequences satisfying the second preset condition are selected from the information gain of each initial shape subsequence in the initial shape subsequence set to obtain an intermediate shape subsequence set; wherein the second preset condition includes that the information gain is not less than a second preset threshold value and / or when the information gain is sorted from large to small, the information gain is ranked within the first second preset number; A shape subsequence that meets a target condition is selected from the intermediate shape subsequence set as a shape subsequence corresponding to the first time series.

5. The method according to claim 2, wherein: The second module includes a first convolutional neural network and a second convolutional neural network, wherein the first convolutional neural network and / or the second convolutional neural network are composed of a one-dimensional convolutional neural network, a batch normalization neural network, and a Gaussian error linear network; The step of inputting the first time series set into the second module of the prediction model to obtain a second feature includes: Inputting each first time series in the first time series set into the first convolutional neural network to obtain a time pattern of the first time series set; The time pattern of the first time series set is input into the second convolutional neural network to obtain the second feature.

6. The method according to any one of claims 1 to 3, further comprising: When the predicted full roll rate does not meet the preset requirements, obtaining at least one process parameter corresponding to a historical spinning product of the same product type as the first spinning product from a plurality of candidate spinning products included in the historical knowledge graph, wherein the historical knowledge graph includes the plurality of candidate spinning products and at least one process parameter corresponding to each of the candidate spinning products; Based on at least one process parameter corresponding to the historical spinning product, the process parameter set is updated, and the operation of obtaining the first time series set corresponding to the process parameter set is returned to be executed until the predicted full roll rate meets the preset requirement.

7. According to the method of any one of claims 1 to 6, after the target equipment is bound to the target process sheet and during the process of producing the first spinning product based on the target process sheet, the target process sheet of the target equipment is updated based on the following method: In the process of generating the first spun product, obtaining a physical inspection result of a product sample of the first spun product being produced by the target device; When the physical inspection result meets the physical inspection requirements, obtaining the target process parameters used by the target equipment to produce the product sample; The target process sheet is updated based on the target process parameters.

8. A device for generating a process sheet for a spinning product, comprising: An acquisition module is used to acquire a first time series set corresponding to a process parameter set, wherein the process parameter set includes a plurality of process parameters, each of which is an equipment parameter used by a target equipment to produce a first spun product, and the first time series set includes a plurality of first time series, each of which represents a change of a corresponding process parameter over time; A prediction module, configured to input the first time series set into a prediction model to obtain a predicted full roll rate of the first spun product; A processing module is used to fill each process parameter in the process parameter set into a preset template to form a target process sheet for the first spinning product when the predicted full roll rate meets the preset requirements.

9. The device according to claim 8, wherein: The prediction module comprises: A first processing unit is used to input the first time series set into a first module in the prediction model to obtain a first feature; the first module is used to capture the features of each first time series; A second processing unit is used to input the first time series set into a second module in the prediction model to obtain a second feature; the second module is used to capture the temporal features of the first time series set as a whole; a fusion unit, configured to fuse the first feature and the second feature to obtain a fused feature; A prediction unit is used to input the fusion feature into the classification head module in the prediction model to obtain the predicted full roll rate of the first spun product.

10. The device according to claim 9, wherein: The first module includes a shape filter and a position encoder; The first processing unit comprises: an extraction subunit, configured to extract shape subsequences corresponding to the first time series from the first time series set by using the shape filter; An input subunit is used to input the shape subsequences corresponding to each of the first time series into the position encoder to obtain the first feature.

11. The device according to claim 10, wherein: The extraction subunit is specifically used for: For each first time series in the first time series set, the following operations are performed respectively: Selecting a preset number of data points in the first time series to obtain an initial perception point set; Determine a reconstruction distance between an initial perception point in the initial perception point set and each other data point in the first time series except the initial perception point, and obtain a reconstruction distance of each other data point; Data points that meet a first preset condition are selected from the reconstruction distances of the other data points, and are added to the initial perception point set to obtain a candidate perception point set; wherein the first preset condition includes that the reconstruction distance is not less than a first preset threshold and / or when the reconstruction distances are sorted from large to small, the reconstruction distance is ranked within a first preset number; Acquire a specified number of candidate perception points from the candidate perception point set to obtain an initial shape subsequence set; wherein the initial shape subsequence set includes at least one initial shape subsequence, and the candidate perception point set includes a plurality of candidate perception points; The initial shape subsequences satisfying the second preset condition are selected from the information gain of each initial shape subsequence in the initial shape subsequence set to obtain an intermediate shape subsequence set; wherein the second preset condition includes that the information gain is not less than a second preset threshold value and / or when the information gain is sorted from large to small, the information gain is ranked within the first second preset number; A shape subsequence that meets a target condition is selected from the intermediate shape subsequence set as a shape subsequence corresponding to the first time series.

12. The device according to claim 9, wherein: The second module includes a first convolutional neural network and a second convolutional neural network, wherein the first convolutional neural network and / or the second convolutional neural network are composed of a one-dimensional convolutional neural network, a batch normalization neural network, and a Gaussian error linear network; The second processing unit comprises: A first acquisition subunit is used to input each first time series in the first time series set into the first convolutional neural network to obtain a time pattern of the first time series set; The second acquisition subunit is used to input the time pattern of the first time series set into the second convolutional neural network to obtain the second feature.

13. The apparatus according to any one of claims 8 to 10, further comprising a first updating module, configured to: When the predicted full roll rate does not meet the preset requirements, at least one process parameter corresponding to a historical spinning product of the same product type as the first spinning product is obtained from a plurality of candidate spinning products included in the historical knowledge graph, wherein: The historical knowledge graph includes the plurality of candidate spinning products and at least one process parameter corresponding to each of the candidate spinning products; Based on at least one process parameter corresponding to the historical spinning product, the process parameter set is updated, and the operation of obtaining the first time series set corresponding to the process parameter set is returned to be executed until the predicted full roll rate meets the preset requirement.

14. The device according to any one of claims 8 to 13, further comprising a second updating module, configured to: after the target device is bound to the target process sheet and during the process of producing the first spinning product based on the target process sheet, update the target process sheet of the target device based on the following method: In the process of generating the first spun product, obtaining a physical inspection result of a product sample of the first spun product being produced by the target device; When the physical inspection result meets the physical inspection requirements, obtaining the target process parameters used by the target equipment to produce the product sample; The target process sheet is updated based on the target process parameters.

15. An electronic device, comprising: at least one processor; as well as a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1 to 7.

16. A non-transitory computer-readable storage medium storing computer instructions, wherein: The computer instructions are used to cause the computer to execute the method according to any one of claims 1-7.

Images