Method for controlling an elasticity imparting process, method for training a control network, and related apparatus.

Abstract

To provide a control method for a texturing process, a training method for a control network, and related apparatuses.SOLUTION: A specific implementation solution includes: sequentially collecting preset control parameters of a plurality of moments in the texturing process to obtain a first control parameter sequence; inputting the first control parameter sequence into a parameter prediction model to predict prediction parameters of a plurality of future moments; selecting a target parameter of a target moment from the prediction parameters of a plurality of moments; building a second control parameter sequence containing the target parameter on the basis of the control parameters before and after the target moment; processing the second control parameter sequence on the basis of an index prediction model to obtain an index prediction result; and inputting the index prediction result into a data generation model in a case where the index prediction result does not meet a desired value to obtain a desired control parameter for the target moment.SELECTED DRAWING: Figure 2

Description

[Technical Field]

[0001] This disclosure relates to the technical field of data processing, and more particularly to a method for controlling an elasticity-granting process, a method for training a control network, and related apparatus. [Background technology]

[0002] In the elasticity-granting (false twist) process, for the convenience of subsequent transportation and management, POY (Pre-Oriented Yarn), abbreviated as POY yarn, is processed into DTY (Draw Textured Yarn), and the DTY yarn is wound onto paper spools to form a DTY yarn package.

[0003] The elasticity imparting process involves many controlled components, and the parameter settings and control of these components have a significant impact on the final quality of the wound yarn package. [Overview of the project]

[0004] This disclosure provides a method for controlling an elasticity-granting process, a method for training a control network, and related apparatus for improving the elasticity-granting process flow.

[0005] According to one aspect of this disclosure, a method for controlling an elasticity imparting process is provided, and the control method is: The process involves sequentially collecting predetermined control parameters at multiple points in time during the elasticity imparting process flow to obtain a first control parameter sequence, The first control parameter sequence is input into a parameter prediction model to predict the parameters at multiple future time points, Selecting target parameters for a target time period from prediction parameters at multiple time periods, Based on control parameters at multiple time points before and after the target time point, a second control parameter sequence including the target parameter is constructed, The process involves processing a second control parameter sequence based on the indicator prediction model to obtain the indicator prediction result, and If the indicator prediction results do not meet the desired values, this includes inputting the indicator prediction results into a data generation model to obtain the desired control parameters for the target time point.

[0006] According to one aspect of this disclosure, a method for training a control network of an elasticity-granting process is provided, wherein the control network includes a parameter prediction subnetwork, an indicator prediction subnetwork, and a data generation subnetwork, and the method is Obtaining a first training sample containing predetermined control parameters at multiple time points in the elasticity imparting process, The first training sample is input into the parameter prediction subnetwork to obtain the parameter prediction values ​​for at least one future time point, The process involves processing parameter prediction values ​​based on the indicator prediction subnetwork to obtain the predicted indicator, and The training loss is determined based on the difference between the predicted indicator and the actual indicator, and the difference between the predicted parameter value and the actual parameter value. Adjusting the parameter prediction subnetwork and the indicator prediction subnetwork based on the training loss, If the training convergence condition is met, the parameter prediction model corresponding to the parameter prediction subnetwork and the index prediction model corresponding to the index prediction subnetwork are obtained. This includes training a data generation subnetwork based on a parameter prediction model and an indicator prediction model to obtain a data generation model, The data generation model is designed to generate desired control parameters for the elasticity enhancer based on the indicator prediction results output from the indicator prediction model.

[0007] According to another aspect of this disclosure, a control device for an elasticity imparting process is provided, the control device is, A collection unit for sequentially collecting predetermined control parameters at multiple time points in the elasticity imparting process to obtain a first control parameter sequence, A parameter prediction unit inputs a first control parameter sequence into a parameter prediction model to predict predict parameters at multiple future time points, A selection unit for selecting target parameters for a target time point from prediction parameters at multiple time points, A construction unit for constructing a second control parameter sequence including the target parameter based on control parameters at multiple points in time before and after the target point in time, An indicator prediction unit for processing a second control parameter sequence based on an indicator prediction model and obtaining indicator prediction results, The system includes a data generation unit that, if the indicator prediction results do not meet the desired values, inputs the indicator prediction results into a data generation model to obtain the desired control parameters for the target time point.

[0008] According to one aspect of this disclosure, a training device for a control network of an elasticity imparting process is provided, the control network including a parameter prediction subnetwork, an indicator prediction subnetwork, and a data generation subnetwork, and the device is An acquisition unit for acquiring a first training sample containing predetermined control parameters at multiple points in time during the elasticity imparting process, The first training sample is input to the parameter prediction subnetwork to obtain the parameter prediction value for at least one future time point, A processing unit for processing parameter prediction values ​​based on an indicator prediction subnetwork and obtaining predicted indicators, A training loss unit is used to determine the training loss based on the difference between the predicted indicator and the actual indicator, and the difference between the predicted parameter value and the actual parameter value. A tuning unit for adjusting parameter prediction subnetworks and indicator prediction subnetworks based on training loss, If the training convergence condition is met, a first generation unit is used to obtain a parameter prediction model corresponding to the parameter prediction subnetwork and an indicator prediction model corresponding to the indicator prediction subnetwork. It includes a second generation unit for training a data generation subnetwork based on a parameter prediction model and an indicator prediction model to obtain a data generation model, The data generation model is designed to generate desired control parameters for the elasticity enhancer based on the indicator prediction results output from the indicator prediction model.

[0009] According to another aspect of this disclosure, an electronic device is provided, and the electronic device is, At least one processor, Includes memory that is communicably connected to at least one processor, The memory stores instructions that can be executed by the at least one processor, and these instructions are executed by the at least one processor to cause the at least one processor to perform a method of any embodiment of the present disclosure.

[0010] According to another aspect of the present disclosure, a non-temporary computer-readable storage medium is provided which stores computer instructions, the computer instructions causing the computer to perform a method according to any embodiment of the present disclosure.

[0011] In other aspects of the present disclosure, a computer program product is provided which, when executed by a processor, implements a method according to any embodiment of the present disclosure.

[0012] In embodiments of this disclosure, predetermined control parameters are collected, future parameters are predicted using a parameter prediction model, target parameters are selected, and a second control parameter sequence is constructed. The second control parameter sequence is processed using an index prediction model to obtain index prediction results, and if the desired values ​​are not met, the parameters in the elasticity imparting process are adjusted by obtaining the desired control parameters using a data generation model.

[0013] The content described in the "Summary of the Invention" section should not be construed as limiting the key points or important features of the embodiments of the present disclosure, nor as limiting the scope of the present disclosure. Other features of the present disclosure will be readily understood from the following description.

[0014] The drawings are provided for a better understanding of the technical solution and do not limit the present disclosure.

Brief Description of the Drawings

[0015] [Figure 1] Figure 1 is a schematic diagram of an elastic imparting machine according to an embodiment of the present disclosure. [Figure 2] Figure 2 is a flowchart of a control method for an elastic imparting process according to an embodiment of the present disclosure. [Figure 3] Figure 3 is a flowchart of a training method for a control network of an elastic imparting process according to an embodiment of the present disclosure. [Figure 4] Figure 4 is a schematic structural diagram of a training parameter prediction sub-network according to an embodiment of the present disclosure. [Figure 5] Figure 5 is a flowchart of obtaining an elastic imparting process architecture chart according to an embodiment of the present disclosure. [Figure 6] Figure 6 is a schematic diagram of an elastic imparting process architecture chart obtained by a second control parameter sequence according to an embodiment of the present disclosure. [Figure 7] Figure 7 is a flowchart of providing a feature module according to an embodiment of the present disclosure. [Figure 8] Figure 8 is a schematic structural diagram of a control device for an elastic imparting process according to an embodiment of the present disclosure. [Figure 9] Figure 9 is a schematic structural diagram of a control network training device for an elastic imparting process according to an embodiment of the present disclosure. [Figure 10] Figure 10 is a block diagram of an electronic device for implementing a control method for an elastic imparting process / a training method for a control network according to an embodiment of the present disclosure. [Modes for carrying out the invention]

[0016] Hereinafter, exemplary embodiments of the present disclosure will be described with reference to the drawings, and various details of the embodiments of the present disclosure will be provided for the sake of ease of understanding, but it should be understood that these are illustrative only. Accordingly, those skilled in the art should be aware that various changes and modifications can be made to the embodiments described herein without departing from the scope of the present disclosure. Similarly, for clarity and brevity, descriptions of known functions and structures will be omitted in the following description.

[0017] The terms "First," "Second," etc., in the specification, claims, and drawings of this application are for distinguishing similar subjects and are not intended to describe a specific order or sequence. It should be understood that such expressions may be interchangeable in appropriate circumstances so that the embodiments of this application described herein may be carried out in an order other than those illustrated or described herein. The embodiments described in the following exemplary examples are not representative of all embodiments conforming to this application. On the contrary, they are merely examples of apparatuses and methods conforming to some aspects of this application, which are described in detail in the appended claims.

[0018] A schematic diagram of an exemplary elasticity-granting machine is shown in Figure 1, and the main components of the elasticity-granting machine include a yarn rack 101, a yarn cutter 102, a first roller 103, a first heating box 104, a cooling plate 105, a false twist member 106, a nozzle 107, a second roller 108, a second heating box 109, a third roller 110, a yarn break detection device 111, and a winding member 112.

[0019] The first roller 103, the second roller 108, and the third roller 110 are designed to ensure that the yarn is processed along a predetermined path. In the elasticity-granting process flow, the speeds of the first roller 103, the second roller 108, and the third roller 110 are harmonized to prevent the yarn from being torn or piled up.

[0020] Depending on the product form needs, the nozzle 107 may be used to process yarn that has been processed via a false-twist member into a network-shaped yarn in order to achieve the desired texture and form for different yarns.

[0021] When producing high-elasticity yarn, the second heating box 109 is not used. When producing medium-elasticity yarn, the temperature of the second heating box 109 may be adjusted to the first predetermined temperature, for example, the first predetermined temperature may be around 140°C. When producing low-elasticity yarn, the temperature of the second heating box 109 may be adjusted to the second predetermined temperature, for example, the second predetermined temperature may be 165-195°C.

[0022] If the thread break detection device 111 detects a thread break, it triggers the thread cutter 102 to cut the thread to avoid accumulation of POY yarn in subsequent flows.

[0023] The elasticity imparting process may include at least one of the following core control parameters: For example, 1) The operating speed, which is the speed of the second roller. 2) The draft ratio, which is the ratio of the speed of the second roller to the speed of the first roller. 3) Secondary overfeed, which is the ratio of the speed difference between the third roller and the second roller to that of the second roller. 4) The false twist ratio, which is the ratio of the disc rotation speed to the operating speed within the false twist member. 5) The rotational speed of the oil wheel, which is the oil supply rotation speed of the oil wheel. Here, the oil wheel is located between the third roller and the winding member and is for supplying oil to the thread. 6) Full winding time, which is the time required for a winding member to complete the full winding of one yarn package. 7) The temperature of the deformed heating box (i.e., the first heating box) is the upper heating box temperature H1. 8) The temperature of the shape-stabilized heating box (i.e., the second heating box) is the temperature of the lower heating box H2. 9) The atmospheric pressure required for the nozzle to strike the network.

[0024] Of course, the parameters listed above are merely illustrative, and the crucial parameters that need to be addressed can be determined during implementation, depending on the specific circumstances.

[0025] The impact of the elasticity imparting process on the quality of DTY (Dynamic Tyre) wound packages is a complex and systematic issue. Inappropriate elasticity imparting process parameters can lead to several defects in the final product. Therefore, a method is needed to optimize the elasticity imparting process to improve product quality and production efficiency, enabling timely detection and resolution of problems during the production process.

[0026] In view of this, the embodiments of the present disclosure provide a method for controlling the elasticity imparting process in order to solve the above problems.

[0027] As shown in Figure 2, the control method for the elasticity imparting process according to the embodiment of this disclosure includes the following:

[0028] S201: Predetermined control parameters at multiple time points in the elasticity imparting process flow are sequentially collected to obtain a first control parameter sequence.

[0029] In the embodiments of this disclosure, predetermined control parameters include operating speed, draft ratio, secondary overfeed, false twist ratio, oil wheel rotation speed, full winding time, upper heating box temperature H1, lower heating box temperature H2, and atmospheric pressure.

[0030] S202: The first control parameter sequence is input into the parameter prediction model to predict the parameters for multiple future time points.

[0031] S203: Select the target parameter for the target time period from the prediction parameters for multiple time periods.

[0032] S204: A second control parameter sequence including the target parameter is constructed based on the control parameters at multiple time points before and after the target time point.

[0033] S205: Process the second control parameter sequence based on the indicator prediction model and obtain the indicator prediction result.

[0034] S206: If the indicator prediction result does not meet the desired value, the indicator prediction result is input into the data generation model to obtain the desired control parameter for the target time point.

[0035] In the embodiments of this disclosure, predetermined control parameters at multiple points in time in the elasticity impartment process flow are collected and input into a parameter prediction model as a first control parameter sequence to predict parameters at multiple future points in time. Then, a target parameter for a target point in time is selected from these predicted parameters. A second control parameter sequence including the target parameter is constructed based on the control parameters at multiple points in time before and after the target point in time. By processing the second control parameter sequence using an indicator prediction model, the indicator prediction result for the future target point in time can be predicted. If the indicator prediction result does not meet the desired value, possible parameter problems in the future can be identified in a timely manner, and the desired control parameter for the future target point in time can be obtained by inputting the indicator prediction result into a data generation model. By making predictions and adjustments to the control parameters, the elasticity impartment process flow is optimized, and the quality of the final product is ensured to meet expectations as much as possible.

[0036] To better understand this, we will explain it from two perspectives: model training and model usage.

[0037] 1. Model Training

[0038] The parameter prediction model, metric prediction model, and data generation model in the embodiments of this disclosure are obtained by training a control network, which includes a parameter prediction subnetwork (for generating the corresponding parameter prediction model after training), a metric prediction subnetwork (for generating the corresponding metric prediction model after training), and a data generation subnetwork (for generating the corresponding data generation model after training).

[0039] Based on this, embodiments of the present disclosure provide a method for training a control network for an elasticity imparting process, which may include the following steps, as shown in Figure 3.

[0040] S301: A first training sample is obtained that includes predetermined control parameters at multiple points in time during the elasticity imparting process.

[0041] For example, data may be collected for key parameters during the actual elasticity imparting process, or multiple predetermined control parameters in the elasticity imparting process may be acquired at the same time point. These predetermined control parameters are key parameters in the elasticity imparting process.

[0042] Centered around time point t, a first training sample corresponding to time point t can be constructed. For example, the first training sample can be constructed by selecting predetermined control parameters corresponding to L time points. For instance, the first training sample can be constructed by selecting predetermined control parameters corresponding to time points t2, t-1, t, t+1, and t+2.

[0043] This allows us to build a corresponding first training sample for each sampling point.

[0044] S302: Input the first training sample into the parameter prediction subnetwork to obtain parameter prediction values ​​for at least one future time point.

[0045] In the embodiments of this disclosure, the Koopman time series model is selected as the parameter prediction subnetwork. The specific training steps are as follows:

[0046] S3021: Disentangle the input first training sample to decompose it into time-varying and time-invariant components.

[0047] For example, if the length of the training sample is T, a Fourier transform is performed on each sequence to obtain the amplitude values ​​corresponding to each frequency in the spectral set S = {0, 1, ..., {T / 2}}. For each frequency, the amplitude values ​​corresponding to that frequency are averaged to obtain the magnitude of the main frequency in the entire sample (the frequency with the highest average amplitude can be considered the main frequency). A set g is obtained by selecting the top frequencies with high average amplitude values, where the proportion is α. α X is the time-invariant component inv Therefore, the remaining component is the time-varying component X. VAR This can be done. The specific steps are as shown in equation (1) below.

[0048]

number

[0049] In equation (1), F represents the FFT (fast Fourier transform), and F -1 represents the inverse fast Fourier transform, Filter() passes only the spectrum corresponding to the input, and X represents the parameters of each sample in the training samples.

[0050] S3022: For the time-invariant component, use the time-invariant Koopman predictor to extract the embedded features of the time-invariant component as the first parameter prediction value.

[0051] As shown in Figure a in Figure 4, input the time-invariant component into the time-invariant Koopman predictor, and map the time-invariant component to the R D latent space by the Encoder encoder, that is, as shown in the following formula (2).

[0052]

Number

[0053] Next, in the latent space, perform a transformation using the learnable matrix K D×D with size R inv , that is,

[0054]

Number

[0055] Finally, map it to R H×C by the decoder, that is,

[0056]

Number

[0057] Y inv is the first parameter prediction value at a future time point.

[0058] S3023: For the time-varying component, use the time-varying Koopman predictor to extract the embedded features and residuals of the time-varying component as the second parameter prediction value.

[0059] As shown in Figure 4b, the time-varying component is input to the time-varying Koopman predictor, and to improve processing efficiency, the input time-varying component is divided into multiple segments of length S (the figure shows an example where it is divided into three segments), and each segment is X j It is labeled as X j This comes from one of the elements in all segments and can be expressed as shown in equation (5) below.

[0060]

number

[0061] In equation (5), X j This shows the time-varying components of each segment that has been divided.

[0062] Learn the transformation of a high-dimensional linear system using a pair of encoders / decoders, i.e.,

[0063]

number

[0064] In equation (5), Z j This shows the corresponding features of the corresponding segment obtained by the corresponding encoder. JPEG0007857447000007.jpg86 is, The new variables obtained by decoding JPEG0007857447000008.jpg86 are shown.

[0065] According to the corresponding features obtained, the time-varying operator is obtained using the data-driven EDMD (Mode Decomposition Algorithm) algorithm, i.e., as shown in equation (7) below.

[0066]

number

[0067] Based on the obtained operators, fitting results can be calculated for power systems observed in the past, namely, as shown in equation (8) below.

[0068]

number

[0069] Furthermore, by extrapolating, we can obtain the predicted result Z for the future. T / S+t It is obtained, that is, as shown in equation (9) below.

[0070]

number

[0071] Finally, by combining these segments, the module's fitting output and sequence prediction can be obtained, as shown in equation (10) below.

[0072]

number

[0073] (10) JPEG0007857447000013.jpg912 is the input data for the next KoopaBlock layer, Y var This is the predicted value of the second parameter at a future point in time.

[0074] The execution order of S3022 and S3023 described above is not restricted, and in some cases, they may be performed simultaneously.

[0075] S303: Process parameter prediction values ​​based on the indicator prediction subnetwork and obtain the predicted indicator. This can be done by including the following steps:

[0076] S3031: If the parameter prediction value is a third parameter sequence constructed from parameters at multiple time points, a specified parameter at a specified time point is selected from the third parameter sequence, and based on the obtained specified parameter, a control parameter sample at multiple time points is constructed based on the time point of that specified parameter, and a fourth control parameter sequence is obtained.

[0077] S3032: Process the fourth control parameter sequence based on the indicator prediction subnetwork to obtain the predicted indicator.

[0078] In some embodiments, a corresponding elasticity-granting process architecture chart can be generated based on the control parameters at each point in time of the fourth control parameter sequence, multiple elasticity-granting process architecture charts can be obtained, and these multiple elasticity-granting process architecture charts can be input into an indicator prediction subnetwork to obtain a predicted indicator.

[0079] Here, the corresponding elasticity-granting process architecture chart is generated by performing the following operations (including steps A1 to A2) on each control parameter at each point in the obtained fourth control parameter sequence.

[0080] Step A1: By performing a normalization operation on each of the multiple subparameters in the control parameter, the subparameters are mapped to a predetermined value interval, and the normalized values ​​of the subparameters are obtained.

[0081] Step A2: By scaling the processing path in the POY yarn elasticity imparting process to the initialization image in an equal proportion (equal scale), an elasticity imparting process architecture chart corresponding to the time point is obtained.

[0082] Here, in the elasticity-granting process architecture chart, the pixel points other than the processing path take their default values, the points corresponding to subparameters in the control parameters on the processing path are set to the normalized value of the subparameter, and the points other than the subparameters in the control parameters on the processing path are set to target values ​​different from the default values.

[0083] The above S3032 can be carried out as the following steps B1 to B4.

[0084] Step B1: Based on the short-term feature extraction module constructed by the attention mechanism in the indicator prediction subnetwork, features are extracted within a third feature extraction range in multiple elasticity-granting process architecture charts to obtain short-term features. Here, the third feature extraction range includes elasticity-granting process architecture charts at n time points centered on the elasticity-granting process architecture chart at the target time point, where n is a positive integer. The short-term feature extraction module uses the elasticity-granting process architecture chart at the target time point as the query feature, and the elasticity-granting process architecture charts at n time points excluding the target time point as key features and value features to obtain short-term features.

[0085] Step B2: Based on the long-term feature extraction module constructed by the attention mechanism in the indicator prediction subnetwork, feature extraction is performed within the fourth feature extraction range in multiple elasticity-granting process architecture charts to obtain long-term features. Here, the fourth feature extraction range includes m elasticity-granting process architecture charts at different time points, centered on the elasticity-granting process architecture chart at the target time point, where m is a positive integer greater than n. The long-term feature extraction module divides the m elasticity-granting process architecture charts at different time points using a sliding window mechanism, determines the pooling information within each sliding window, constructs the key features and value features necessary for the attention mechanism based on the pooling information of multiple sliding windows, and obtains long-term features based on the key features, value features, and query features.

[0086] Step B3: Use the fusion module to perform a fusion operation on long-term and short-term features and obtain fused features.

[0087] Step B4: Process fused features based on the prediction module of the metric prediction subnetwork to obtain the predicted metric.

[0088] To make the predicted metric results obtained by the metric prediction subnetwork, which processes the parameter prediction values, more accurate, the metric prediction subnetwork further includes a supplementary feature extraction module, which can obtain supplementary features by performing the following operations based on the obtained parameter prediction values.

[0089] Step C1: The obtained fourth control parameter sequence is analyzed to obtain the cumulative difference of the parameters at each time point in the fourth control parameter sequence relative to the first time point in the fourth control parameter sequence, thereby obtaining the cumulative difference sequence. The parameter at the first time point is the first control parameter in the fourth control parameter sequence.

[0090] Step C2: Based on the cumulative difference sequence, determine the time length relative to the first time point in the fourth control parameter sequence for the time point when the cumulative difference is greater than a predetermined threshold, and obtain the time length sequence.

[0091] Step C3: Feature extraction is performed on the obtained fourth control parameter sequence, cumulative difference sequence, and time length sequence to obtain supplementary features.

[0092] Step C4: Use the fusion module to perform a fusion operation on long-term features, short-term features, and supplementary features to obtain fused features.

[0093] S304: Determine the training loss based on the difference between the predicted indicator and the actual indicator, and the difference between the predicted parameter value and the actual parameter value.

[0094] S305: Adjust the parameter prediction subnetwork and the indicator prediction subnetwork based on the training loss.

[0095] S306: If the training convergence condition is met, obtain the parameter prediction model corresponding to the parameter prediction subnetwork and the indicator prediction model corresponding to the indicator prediction subnetwork.

[0096] In the embodiments of this disclosure, the mean squared error can be used as the loss function to calculate the difference between the predicted indicator and the actual indicator, and the difference between the predicted parameter values ​​and the actual parameter values, thereby allowing the parameter prediction subnetwork and the indicator prediction subnetwork to be adjusted. Finally, the model is trained through iterations, and the model's performance is evaluated using a validation set or a test set. Under conditions that satisfy the requirements, the optimized parameter prediction subnetwork is confirmed as the parameter prediction model, and the optimized indicator prediction subnetwork is confirmed as the indicator prediction model.

[0097] S307: Based on the parameter prediction model and the indicator prediction model, a data generation subnetwork is trained to obtain the data generation model, which generates the desired control parameters for the elasticity imparter based on the indicator prediction results output from the indicator prediction model. Here, the generation subnetwork includes a training-wait generator and a training-wait discriminator, and based on the parameter prediction model and the indicator prediction model, the data generation subnetwork is trained to obtain the data generation model.

[0098] To ensure the accuracy of the resulting data generation model, it is necessary to train the data generation subnetwork multiple times, with each training session performing the following operations.

[0099] Step D1: Input the true control parameters into the training-waiting classifier and obtain the first classifier's decision regarding the true control parameters.

[0100] Step D2: Generate false control parameters using the training wait generator.

[0101] Step D3: Input false control parameters into the training-wait classifier and obtain a second classification result.

[0102] Step D4: Adjust the model parameters of the classifier awaiting training based on the first and second classification results.

[0103] For example, a training-wait generator is used to generate false control parameters, and a truth / false dataset is constructed by combining the generated false control parameters with the true control parameters. Based on the constructed truth / false dataset, a training-wait discriminator is trained without changing the generator parameters. For example, the training-wait discriminator is trained to determine whether a single random training sample is a true control parameter or a false sample generated by the training-wait generator. The true control parameters are input into the training-wait discriminator, and the resulting discrimination result is taken as the true result. The obtained false control parameters are input into the training-wait discriminator, and the resulting discrimination result is taken as the false result. The training of the training-wait discriminator is completed by referring to the true and false results.

[0104] Step D5: Fix the model parameters of the training queue discriminator and train the training queue generator.

[0105] After completing the training of the training-wait discriminator, the training-wait generator is connected in series with the trained training-wait discriminator, and the training-wait generator is trained while keeping the parameters of the training-wait discriminator fixed. The training-wait generator is given the prediction index output from the index prediction model, and the control parameters generated by the training-wait generator are obtained. These generated control parameters are input to the training-wait discriminator for judgment, and a loss value is generated as the target of processing in the loss function based on the recognition result of the training-wait discriminator. The parameters of the training-wait generator are then updated based on this loss value.

[0106] After completing the training of the training queue generator, the training queue discriminator is trained again, keeping the training queue generator's parameters fixed. This process is repeated until the training queue generator can generate sufficiently realistic data and the training queue discriminator cannot accurately distinguish between true and false data. This yields a data generation model.

[0107] In the embodiments of this disclosure, the training generator and training discriminant process the predicted parameters generated by the parameter prediction model and the predicted index results generated by the index prediction model to obtain a data generation model and provide efficient optimization efficiency for parameters in the elasticity-granting process flow.

[0108] In the embodiments of this disclosure, a parameter prediction subnetwork is trained with a first training sample to obtain parameter prediction values. An index prediction subnetwork is trained based on the obtained parameter prediction values ​​to obtain index parameters. A parameter prediction model and an index prediction model are obtained by determining the training loss and adjusting the parameter prediction subnetwork and the index prediction subnetwork. Based on this, a data generation subnetwork can be trained to obtain a data generation model. By training the subnetworks to obtain the final prediction model, the optimization efficiency for parameters in the elasticity-granting flow is improved.

[0109] 2. Using the Model

[0110] The desired DTY yarn is obtained when the POY yarn passes through the yarn processing path and the operation of the elastic-granting process is completed. The entire yarn path is designed based on the elastic-granting process of the elastic-granting machine. In addition to the processing of corresponding critical parts throughout the entire process for the yarn, several mechanical properties of the elastic-granting machine and the processing of the POY yarn in the yarn path also affect the quality of the final yarn. In order to uncover the impact of potential hidden features on the quality of the DTY yarn, embodiments of this disclosure can be carried out by processing a second control parameter sequence based on an index prediction model and obtaining index prediction results, which involves generating a corresponding elastic-granting process architecture chart based on the control parameters at each point in the second control parameter sequence to obtain multiple elastic-granting process architecture charts, and then inputting the multiple elastic-granting process architecture charts into the index prediction model to obtain index prediction results.

[0111] In the embodiments of this disclosure, the elasticity imparting process architecture chart can represent the yarn processing path in the yarn elasticity imparting machine and the essential parameters of key parts. This allows for the uncovering of potential features and relationships by linking predicted control data with the elasticity imparting process path in the elasticity imparting machine and converting it into a diagrammatic format, thereby facilitating the prediction of quality at future points in time.

[0112] During implementation, the generation of the elasticity-granting process architecture chart is the same as during model training. As shown in Figure 5, the corresponding elasticity-granting process architecture chart is generated by performing the operations shown in Figure 5 for each control parameter at each time point in the second control parameter sequence.

[0113] S501: Perform a normalization operation on each of the multiple subparameters in the control parameter, map the subparameters to a predetermined value interval, and obtain the normalized value of the subparameter.

[0114] For example, the range of data for each sub-parameter can be determined by calculating the maximum and minimum values ​​for each sub-parameter.

[0115] The normalization operation on subparameters can be calculated using the following formula.

[0116]

number

[0117] In equation (11), y represents the new value obtained by transforming each data point in the subparameter from its original range to the range [0,1], x represents each subparameter, min represents the minimum value in the subparameter, and max represents the maximum value in the subparameter.

[0118] To reduce the computational complexity in subsequent training processes, the newly calculated values ​​are mapped to a predetermined value interval. In this embodiment, the predetermined value interval is 0 to 255. Therefore, by multiplying the normalized value by 255 and then adding 0, a value within the range of 0 to 255 can be obtained.

[0119] S502: The processing path in the POY yarn elasticity imparting process is scaled proportionally to an initialization image to obtain an elasticity imparting process architecture chart corresponding to the time point. Here, the values ​​of pixel points other than the processing path in the elasticity imparting process architecture chart take the default value, points corresponding to subparameters in the control parameters on the processing path are set to the normalized value of the subparameter, and points other than the subparameters in the control parameters on the processing path are set to target values ​​different from the default value.

[0120] In the embodiments of this disclosure, by setting the default value to 255 and using this as the value of pixel points other than the processing path in the initialization image, setting the target value to 0 and using this as the value of points other than sub-parameters in the control parameters on the processing path, and using the remaining values ​​as the values ​​of points corresponding to sub-parameters in the control parameters on the processing path, an elasticity imparting process architecture chart as shown in Figure 6 can be obtained. The entire elasticity imparting process flow and the corresponding members may be understood as being scaled to the initial image, and the values ​​of important parts in the figure can represent the process parameters of those parts.

[0121] In the embodiments of this disclosure, normalized values ​​are obtained by mapping multiple subparameters in a second control parameter sequence to a predetermined value interval through a normalization operation. Subsequently, the processing path in the POY yarn elasticity imparting process is scaled proportionally to an initialization image. Pixel points in the initialization image other than the processing path are set to default values, points on the processing path corresponding to subparameters in the control parameters are set to the normalized value of the subparameter, and points other than the subparameters are set to target values ​​different from the default values. In this way, an elasticity imparting process architecture chart is obtained. Converting real-time control parameters into a graph representation provides a good data basis for uncovering potential relationships and features by subsequent indicator prediction models.

[0122] In some embodiments, to better uncover features and improve the accuracy of indicator prediction, multiple elasticity-granting process architecture charts can be input into the indicator prediction model, and indicator prediction results can be obtained, specifically as shown in Figure 7.

[0123] S701: Based on a short-term feature extraction module constructed by an attention mechanism in the indicator prediction model, features are extracted within a first feature extraction range in multiple elasticity-granting process architecture charts to obtain short-term features. Here, the first feature extraction range includes elasticity-granting process architecture charts at n time points centered on the elasticity-granting process architecture chart at the target time point, where n is a positive integer. The short-term feature extraction module uses the elasticity-granting process architecture chart at the target time point as the query feature, and the elasticity-granting process architecture charts at n time points excluding the target time point as key features and value features to obtain short-term features.

[0124] For example, if there is a series of elasticity-granting process architecture charts, each figure represents the process state at a specific point in time. In embodiments of this disclosure, it is desirable to predict the process state at a future point in time and extract short-term features based on attention mechanisms in historical data.

[0125] Therefore, first, we determine the first feature extraction range, that is, the elasticity-granting process architecture chart for n time points centered on the target time point. If we select five time points as the first feature extraction range, we select two time points before and two after the current time point (target time point).

[0126] Next, a short-term feature extraction module is used to extract features. This module uses the elasticity-granting process architecture chart at the target time point as the query feature, and the elasticity-granting process architecture charts at the other four time points as the key feature and value feature, respectively. By calculating the similarity between these features, the model can capture the relationships and trends between different time points.

[0127] S702: Based on a long-term feature extraction module constructed by an attention mechanism in an indicator prediction model, features are extracted within a second feature extraction range in multiple elasticity-granting process architecture charts to obtain long-term features. Here, the second feature extraction range includes m elasticity-granting process architecture charts at different time points, centered on the elasticity-granting process architecture chart at the target time point, where m is a positive integer greater than n. The long-term feature extraction module divides the m elasticity-granting process architecture charts at different time points using a sliding window mechanism, determines pooling information within each sliding window, constructs key features and value features necessary for the attention mechanism based on the pooling information of multiple sliding windows, and obtains long-term features based on the key features, value features, and query features.

[0128] For example, consider a process architecture chart sequence with elasticity, containing 100 time points, where each time point represents a different process state.

[0129] Therefore, first, we determine the second feature extraction range, that is, the elasticity-granting process architecture chart for m time points centered on the target time point. If we select 20 time points as the second feature extraction range, we select 10 time points before and 10 time points after the current time point (target time point).

[0130] Next, feature extraction is performed using a long-term feature extraction module. This module divides the elasticity-granting process architecture chart at the 20 time points using a sliding window mechanism and determines the pooling information within each sliding window. For example, one window size may be set to 5, and then the window may be slid across these 20 time points. Also, assuming that the windows may or may not overlap, pooling information for at least four windows can be obtained. In each window, the model constructs key features and value features based on the attention mechanism.

[0131] S703: After obtaining short-term and long-term features, a fusion operation is performed on the long-term and short-term features using a fusion module to obtain fused features.

[0132] S704: Process fused features based on the prediction module and obtain indicator prediction results.

[0133] In the embodiments of this disclosure, a short-term feature extraction module extracts short-term features from multiple elasticity-granting process architecture charts, and a long-term feature extraction module extracts long-term features from multiple elasticity-granting processes. The obtained short-term and long-term features are then fused as input parameters for a prediction module to obtain indicator prediction results. By extracting and fusion of long-term and short-term features, the characteristics of the data can be captured more comprehensively when processing data parameters with different time scales. This improves the performance and effectiveness of predictions in the elasticity-granting process flow of the prediction model.

[0134] Furthermore, in some embodiments, in order to make the metric prediction results obtained by the prediction model more accurate, the metric prediction model further includes a supplementary feature extraction module, and supplementary features are obtained by performing the following operations based on the supplementary feature extraction module.

[0135] Specifically, the supplementary feature extraction module analyzes the data for the second control parameter sequence, obtains the cumulative difference of the parameters at each time point in the second control parameter sequence relative to the first time point in the second control parameter sequence, obtains a cumulative difference sequence, determines the time length relative to the first time point for times when the cumulative difference is greater than a predetermined threshold based on the cumulative difference sequence, obtains a time length sequence, and extracts features from the second control parameter sequence, the cumulative difference sequence, and the time length sequence to obtain supplementary features.

[0136] For example, there is a second control parameter sequence containing 10 time points, where each time point represents a different control parameter value.

[0137] Therefore, it is first necessary to determine the first time point as the reference time. In the embodiments of this disclosure, the first time point is selected as the first time point.

[0138] Next, we analyze the data for the second control parameter sequence and calculate the cumulative difference of the parameter at each time point relative to the first time point. For example, if the control parameter value at the first time point is 10 and the control parameter value at the second time point is 12, the cumulative difference of the parameter at this time point will be 2. By analogy, a new cumulative difference sequence can be obtained.

[0139] Next, based on the cumulative difference sequence, the time length relative to the first time point for each time point where the cumulative difference is greater than a predetermined threshold is determined. For example, if the predetermined threshold is 3, all time points where the cumulative difference is greater than 3 are found, and the time length between them and the first time point is calculated. This yields a new time length sequence.

[0140] Finally, the second control parameter sequence, cumulative difference sequence, and time length sequence are input into the feature extraction module to obtain supplementary features.

[0141] With supplementary features already present, a fusion module is used to perform a fusion operation on long-term features, short-term features, and supplementary features to obtain fused features.

[0142] In the embodiments of this disclosure, extracting supplementary features allows for obtaining more information about the second control parameter sequence, which helps the model better understand and predict future elasticity-granting process architecture charts. Further optimization of the prediction model by adding extraction modules makes the prediction model results more accurate.

[0143] In the embodiments of this disclosure, the indicator prediction results for the elasticity imparting process include at least one of the following: grade of the wound yarn package, pass rate, full package rate (the degree to which the yarn is wound in the wound yarn package), and dye uniformity.

[0144] In the embodiments of this disclosure, by predicting key indicators such as the grade, acceptance rate, full winding rate, and dye uniformity of the wound yarn package, problems can be identified in a timely manner and addressed to improve them, thereby reducing the defect rate and lowering production costs.

[0145] Based on the same technical concept, embodiments of this disclosure provide a control device for an elasticity imparting process, as shown in Figure 8, A collection unit 801 for sequentially collecting predetermined control parameters at multiple time points in the elasticity imparting process flow to obtain a first control parameter sequence, A parameter prediction unit 802 inputs a first control parameter sequence into a parameter prediction model to predict predict parameters at multiple future time points, A selection unit 803 for selecting target parameters for a target time point from prediction parameters at multiple time points, A construction unit 804 for constructing a second control parameter sequence including the target parameter based on control parameters at multiple points in time before and after the target point, An indicator prediction unit 805 processes a second control parameter sequence based on an indicator prediction model and obtains indicator prediction results, The system includes a data generation unit 806 for inputting the indicator prediction results into a data generation model to obtain desired control parameters for a given time point if the indicator prediction results do not meet the desired values.

[0146] In some embodiments, the indicator prediction unit is: A generation subunit for generating a corresponding elasticity-granting process architecture chart based on the control parameters at each point in time of the second control parameter sequence, and for obtaining multiple elasticity-granting process architecture charts, It includes a prediction subunit for inputting multiple elasticity-granting process architecture charts into an indicator prediction model and obtaining indicator prediction results.

[0147] In some embodiments, the generated subunit is specifically, For each control parameter at each point in the second control parameter sequence, the following operations are performed to generate the corresponding elasticity-granting process architecture chart: The operation in question is, This involves performing a normalization operation on multiple sub-parameters in a control parameter, mapping the sub-parameters to a predetermined value interval, and obtaining the normalized values ​​of the sub-parameters. This includes scaling the processing path in the POY yarn elasticity imparting process proportionally to the initial image to obtain an elasticity imparting process architecture chart corresponding to the time point, Here, in the elasticity-granting process architecture chart, the pixel points other than the processing path take their default values, the points corresponding to subparameters in the control parameters on the processing path are set to the normalized value of the subparameter, and the points other than the subparameters in the control parameters on the processing path are set to target values ​​different from the default values.

[0148] In some embodiments, the prediction subunit is used specifically to perform the following operations: The operation in question is, Based on a short-term feature extraction module constructed using an attention mechanism in an indicator prediction model, features are extracted within a first feature extraction range in multiple elasticity-granting process architecture charts to obtain short-term features. Here, the first feature extraction range includes elasticity-granting process architecture charts at n time points centered on the elasticity-granting process architecture chart at the target time point, where n is a positive integer. The short-term feature extraction module uses the elasticity-granting process architecture chart at the target time point as the query feature, and the elasticity-granting process architecture charts at n time points excluding the target time point as key features and value features to obtain short-term features. Based on a long-term feature extraction module constructed by an attention mechanism in an indicator prediction model, features are extracted within a second feature extraction range in multiple elasticity-granting process architecture charts to obtain long-term features. Here, the second feature extraction range includes m elasticity-granting process architecture charts at different time points, centered on the elasticity-granting process architecture chart at the target time point, where m is a positive integer greater than n. The long-term feature extraction module divides the m elasticity-granting process architecture charts at different time points using a sliding window mechanism, determines pooling information within each sliding window, constructs key features and value features necessary for the attention mechanism based on the pooling information of multiple sliding windows, and obtains long-term features based on the key features, value features, and query features. Using a fusion module, a fusion operation is performed on long-term and short-term features to obtain fused features. This includes processing fused features based on the prediction module in the indicator prediction model to obtain indicator prediction results.

[0149] In some embodiments, the indicator prediction model further includes a supplementary feature extraction module, and the prediction subunit is further used to perform the following processing: The process in question is, Based on the supplementary feature extraction module, perform the following operations to obtain supplementary features, This includes performing a fusion operation on long-term features, short-term features, and supplementary features using a fusion module to obtain fused features. The operation in question is, The data is analyzed for the second control parameter sequence, the cumulative difference of the parameters at each time point in the second control parameter sequence compared to the first time point in the second control parameter sequence is obtained, and the cumulative difference sequence is obtained. Based on the cumulative difference sequence, the time length relative to the first time point for each point in time where the cumulative difference is greater than a predetermined threshold is determined, and a time length sequence is obtained. This includes extracting features from a second control parameter sequence, a cumulative difference sequence, and a time length sequence, and obtaining supplementary features.

[0150] In some embodiments, the indicator prediction results include at least one of the following: grade of the yarn package, pass rate, full winding rate, and dye uniformity.

[0151] Embodiments of this disclosure provide a training device for a control network of an elasticity imparting process, the control network including a parameter prediction subnetwork, an indicator prediction subnetwork, and a data generation subnetwork, as shown in Figure 9. An acquisition unit 901 for acquiring a first training sample including predetermined control parameters at multiple points in time during the elasticity imparting process, Input unit 902 to input the first training sample into the parameter prediction subnetwork to obtain parameter prediction values ​​for at least one future time point, A processing unit 903 for processing parameter prediction values ​​based on an indicator prediction subnetwork and obtaining a predicted indicator, A training loss unit 904 for determining the training loss based on the difference between the predicted indicator and the actual indicator, and the difference between the predicted parameter value and the actual parameter value, A training loss unit 904 for adjusting the parameter prediction subnetwork and the indicator prediction subnetwork based on the training loss, If the training convergence condition is met, a first generation unit 906 is used to obtain a parameter prediction model corresponding to the parameter prediction subnetwork and an indicator prediction model corresponding to the indicator prediction subnetwork. It includes a second generation unit 907 for training a data generation subnetwork based on a parameter prediction model and an indicator prediction model to obtain a data generation model, The data generation model is designed to generate desired control parameters for the elasticity enhancer based on the indicator prediction results output from the indicator prediction model.

[0152] In some embodiments, the data generation subnetwork includes a training queue generator and a training queue discriminator, and the second generation unit is specifically used to perform multiple training runs on the data generation network, with each training run performing the following operations: The aforementioned operation is, The true control parameters are input to the training-waiting classifier, and a first classification result for the true control parameters is obtained by the classifier. Using a training queue generator to generate false control parameters, Inputting false control parameters into the training-waiting classifier and obtaining a second classification result, Adjust the model parameters of the training-waiting classifier based on the first and second classification results, This includes fixing the model parameters of the training queue discriminator and training the training queue generator.

[0153] A description of the specific functions and examples of each module, submodule / unit of the apparatus according to the embodiments of this disclosure can be found in the relevant descriptions of the corresponding steps in the embodiments of the method described above, and is therefore omitted here.

[0154] In the proposed technology disclosed herein, the acquisition, storage, and use of users' personal information will comply with the provisions of applicable laws and regulations and will not violate public order and morals.

[0155] Figure 10 is a block diagram of the configuration of an electronic device according to one embodiment of the present disclosure. As shown in Figure 10, the electronic device includes a memory 1010 and a processor 1020, the memory 1010 storing a computer program that can be executed by the processor 1020. The number of memories 1010 and processors 1020 may be one or more. The memory 1010 may store one or more computer programs. When the one or more computer programs are executed by the electronic device, the electronic device can perform the method provided in the embodiment of the above method. The electronic device may further include a communication interface 1030 for communicating with external devices and exchanging and transmitting data.

[0156] If the memory 1010, processor 1020, and communication interface 1030 are separate components, then the memory 1010, processor 1020, and communication interface 1030 are connected to each other via a bus and can communicate with each other. This bus may be an ISA (Industry Standard Architecture) bus, a PCI (Peripheral Component Interconnect) bus, an EISA (Extended Industry Standard Architecture) bus, etc. This bus can be divided into an address bus, a data bus, a control bus, etc. For convenience, Figure 10 shows only one thick line, but this does not mean that there is only one bus or only one type of bus.

[0157] Selectively, as a concrete implementation, if the memory 1010, processor 1020, and communication interface 1030 are integrated onto a single chip, the memory 1010, processor 1020, and communication interface 1030 can communicate with each other via an internal interface.

[0158] The processor may be a Central Processing Unit (CPU), a general-purpose processor, a Digital Signal Processing (DSP), an Application Specific Integrated Circuit (ASIC), a Field Programmable Gate Array (FPGA), or other programmable logic devices, discrete gates, transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any ordinary processor. Furthermore, the processor may be a processor capable of supporting an Advanced RISC Machine (ARM) architecture.

[0159] Furthermore, the memory may selectively include read-only memory and random access memory, or non-volatile random access memory. The memory may be volatile memory or non-volatile memory, or may include both volatile and non-volatile memory. Non-volatile memory may include ROM (Read-Only Memory), Programmable ROM (PROM), Erasable PROM (EPROM), Electrically EPROM (EEPROM), or flash memory. Volatile memory may include Random Access Memory (RAM) used as an external cache. The above description is illustrative and not restrictive. Many forms of RAM are available. For example, static random access memory (Static RAM, SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (Synchronous DRAM, SDRAM), double data rate synchronous dynamic random access memory (Double Data Date SDRAM, DDR SDRAM), enhanced synchronous dynamic random access memory (Enhanced SDRAM, ESDRAM), synchlink dynamic random access memory (Synchlink DRAM, SLDRAM), and direct memory bus random access memory (Direct RAM BUS RAM, DR RAM) may be used.

[0160] In the embodiments described above, all or part of the embodiments may be implemented by software, hardware, firmware, or any combination thereof. If implemented by software, all or part of the embodiments may be implemented 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, all or part of the flows or functions described in the embodiments of this disclosure are generated. The computer may be a general-purpose computer, a dedicated computer, a computer network, or other programmable device. The computer instructions may be stored on a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions may be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via a wired connection (e.g., coaxial cable, optical fiber, digital subscriber line, DSL) or wireless connection (e.g., infrared, Bluetooth®, microwave, etc.). The computer-readable storage medium may be any available medium accessible by a computer, or it may be a data storage device such as a server or data center that includes one or more available media. The available media may be magnetic media (e.g., floppy disks, hard disks, or magnetic tapes), optical media (e.g., Digital Versatile Discs (DVDs)), semiconductor media (e.g., Solid State Disks (SSDs)), etc. The computer-readable storage medium according to this disclosure may also be a non-volatile storage medium, in other words, a non-temporary storage medium.

[0161] Those skilled in the art will understand that all or part of the steps for realizing the above embodiment may be completed by hardware, or by a program that instructs the relevant hardware, and that the program may be stored in a computer-readable storage medium, the storage medium may be a read-only memory, a magnetic disk, or an optical disk, etc.

[0162] In the descriptions of the embodiments of this disclosure, the terms “one embodiment,” “several embodiments,” “example,” “specific example,” or “several examples” mean that the specific features, structures, materials, or characteristics described in relation to such embodiment or example are included in at least one embodiment or example of this disclosure. Furthermore, the specific features, structures, materials, or characteristics described may be combined in an appropriate manner in any one or more embodiments or examples. Furthermore, a person skilled in the art may combine different embodiments or examples and features in different embodiments or examples described herein, provided that they are not inconsistent.

[0163] In the description of the embodiments of this disclosure, unless otherwise specified, " / " means "or," for example, "A / B" can represent "A" or "B." The "and / or" statements in this specification are merely related relationships that describe related subjects and mean that there are three possible relationships, for example, "A and / or B" can indicate three situations: "A" exists alone, "A" and "B" exist together, and "B" exists alone.

[0164] In the description of the embodiments of this disclosure, the terms “first” and “second” are for distinction purposes only and should not be understood to indicate or imply relative importance or the number of designated constituent elements. Thus, features limited by “first” and “second” may explicitly or implicitly include one or more such features. In the description of the embodiments of this disclosure, unless otherwise specified, “multiple” means two or more.

[0165] The foregoing are merely illustrative examples of the Disclosure and do not limit the Disclosure. Any modifications, equivalent substitutions, or improvements made to the spirit and principles of the Disclosure should be included within the scope of the claims of the Disclosure.

Claims

1. A method for controlling the process of imparting elasticity, The process involves sequentially collecting predetermined control parameters at multiple points in time during the elasticity imparting process flow to obtain a first control parameter sequence, The first control parameter sequence is input into a parameter prediction model to predict the parameters at multiple future time points, Selecting the target parameter for the target time from the prediction parameters of the aforementioned multiple time points, Based on control parameters at multiple time points before and after the aforementioned target time point, a second control parameter sequence including the aforementioned target parameter is constructed. The process involves processing the second control parameter sequence based on the indicator prediction model and obtaining the indicator prediction result. If the indicator prediction result does not meet the desired value, the indicator prediction result is input into the data generation model to obtain the desired control parameter for the target time point, The aforementioned indicator prediction results include at least one of the grade, pass rate, full winding rate, and dyeing uniformity of the wound yarn package, and are a method for controlling the elasticity imparting process.

2. Processing the second control parameter sequence based on the aforementioned indicator prediction model and obtaining the indicator prediction result is: Based on the control parameters at each point in time of the second control parameter sequence, a corresponding elasticity-granting process architecture chart is generated, and multiple elasticity-granting process architecture charts are obtained. The method according to claim 1, further comprising inputting the plurality of elasticity-granting process architecture charts into the indicator prediction model to obtain the indicator prediction results.

3. Based on the control parameters at each point in time of the second control parameter sequence, a corresponding elasticity-granting process architecture chart is generated, and multiple elasticity-granting process architecture charts are obtained. This includes performing the following operations on each control parameter at each point in the second control parameter sequence to generate a corresponding elasticity-granting process architecture chart: The aforementioned operation is, The process involves performing a normalization operation on each of the multiple sub-parameters in the control parameter, mapping the sub-parameters to a predetermined value interval, and obtaining the normalized values ​​of the sub-parameters. This includes scaling the processing path in the POY yarn elasticity imparting process proportionally to the initialization image, and obtaining an elasticity imparting process architecture chart corresponding to the said time point. The method according to claim 2, wherein the values ​​of pixel points other than the processing path in the elasticity imparting process architecture chart are set to default values, the points on the processing path corresponding to the subparameters in the control parameters are set to the normalized value of the subparameters, and the points on the processing path other than the subparameters in the control parameters are set to target values ​​different from the default values.

4. Inputting the aforementioned multiple elasticity-granting process architecture charts into the indicator prediction model to obtain the indicator prediction results is, Based on a short-term feature extraction module constructed by the attention mechanism in the indicator prediction model, features are extracted within a first feature extraction range in the plurality of elasticity-granting process architecture charts to obtain short-term features, wherein the first feature extraction range includes elasticity-granting process architecture charts at n time points centered on the elasticity-granting process architecture chart at the target time point, where n is a positive integer, and the short-term feature extraction module uses the elasticity-granting process architecture chart at the target time point as a query feature, and the elasticity-granting process architecture charts at the n time points excluding the target time point as key features and value features, to obtain the short-term features, and Based on a long-term feature extraction module constructed by the attention mechanism in the indicator prediction model, features are extracted within a second feature extraction range in the plurality of elasticity-granting process architecture charts to obtain long-term features, wherein the second feature extraction range includes m elasticity-granting process architecture charts at various time points centered on the elasticity-granting process architecture chart at the target time point, where m is a positive integer greater than n, and the long-term feature extraction module divides the m elasticity-granting process architecture charts at various time points using a sliding window mechanism, determines pooling information within each sliding window, constructs key features and value features necessary for the attention mechanism based on the pooling information of the plurality of sliding windows, and obtains the long-term features based on the key features, value features, and query features. Using a fusion module, a fusion operation is performed on the long-term features and the short-term features to obtain the fused features. The process involves processing the fused features based on the prediction module in the indicator prediction model to obtain the indicator prediction result, The method according to claim 2, including the method described in claim 2.

5. The aforementioned indicator prediction model further includes a supplementary feature extraction module, The above method further, The following operations are performed based on the aforementioned supplementary feature extraction module to obtain supplementary features, The aforementioned operation is, The data is analyzed for the second control parameter sequence, and the cumulative difference of the parameters at each time point in the second control parameter sequence relative to the first time point in the second control parameter sequence is obtained to obtain a cumulative difference sequence. Based on the cumulative difference sequence, the time length relative to the first time point at which the cumulative difference is greater than a predetermined threshold is determined, and a time length sequence is obtained. Features are extracted from the second control parameter sequence, the cumulative difference sequence, and the time length sequence, and the supplementary features are obtained. Includes, Performing a fusion operation on the long-term features and the short-term features using the fusion module to obtain fused features is: The method according to claim 4, comprising performing a fusion operation on the long-term features, the short-term features and the supplementary features using a fusion module to obtain the fused features.

6. A method for training a control network for an elasticity-granting process, The control network includes a parameter prediction subnetwork, an indicator prediction subnetwork, and a data generation subnetwork. The aforementioned method, Obtaining a first training sample containing predetermined control parameters at multiple time points in the elasticity imparting process, The first training sample is input to the parameter prediction subnetwork to obtain the parameter prediction value for at least one future time point, The process involves processing the parameter prediction values ​​based on the aforementioned indicator prediction subnetwork to obtain the predicted indicator, The training loss is determined based on the difference between the predicted indicator and the actual indicator, and the difference between the predicted parameter value and the actual parameter value. Adjusting the parameter prediction subnetwork and the indicator prediction subnetwork based on the training loss, If the training convergence condition is met, the parameter prediction model corresponding to the parameter prediction subnetwork and the index prediction model corresponding to the index prediction subnetwork are obtained. Based on the parameter prediction model and the indicator prediction model, the data generation subnetwork is trained to obtain a data generation model. Includes, The data generation model is for generating desired control parameters for the elasticity imparting machine based on the index prediction results output from the index prediction model. A method for training a control network for an elasticity imparting process, wherein the indicator prediction results include at least one of the grade, pass rate, full winding rate, and dye uniformity of the wound yarn package.

7. The aforementioned data generation subnetwork includes a training queue generator and a training queue discriminator, To obtain a data generation model by training the data generation subnetwork based on the parameter prediction model and the indicator prediction model, This includes performing multiple training sessions on the aforementioned data generation subnetwork. Each of the aforementioned training sessions will involve performing the following operations: The aforementioned operation is, The true control parameter is input to the training-waiting discriminator, and a first discrimination result for the true control parameter is obtained by the training-waiting discriminator. The training wait generator is used to generate false control parameters, The false control parameters are input to the training-waiting classifier to obtain a second classification result. Adjusting the model parameters of the training-waiting classifier based on the first and second classification results, The model parameters of the aforementioned training wait discriminator are fixed, and the aforementioned training wait generator is trained. The method according to claim 6, including the method described in claim 6.

8. A control device for the elasticity imparting process, A collection unit for sequentially collecting predetermined control parameters at multiple time points in the elasticity imparting process flow to obtain a first control parameter sequence, A parameter prediction unit for inputting the first control parameter sequence into a parameter prediction model to predict predict parameters at multiple future time points, A selection unit for selecting a target parameter for a target time from the multiple prediction parameters at the aforementioned time points, A construction unit for constructing a second control parameter sequence including the target parameter based on control parameters at multiple time points before and after the target time point, An indicator prediction unit for processing the second control parameter sequence based on an indicator prediction model and obtaining indicator prediction results, If the indicator prediction result does not meet the desired value, the data generation unit inputs the indicator prediction result into a data generation model to obtain the desired control parameter for the target time point, and The indicator prediction results include at least one of the grade, pass rate, full winding rate, and dyeing uniformity of the wound yarn package, and are control devices for an elasticity imparting process.

9. The aforementioned indicator prediction unit is A generation subunit for generating a corresponding elasticity-granting process architecture chart based on the control parameters at each point in time of the second control parameter sequence, and for obtaining multiple elasticity-granting process architecture charts, A prediction subunit for inputting the multiple elasticity-granting process architecture charts into the indicator prediction model and obtaining the indicator prediction results, The apparatus according to claim 8, including the apparatus described in claim 8.

10. The generation subunit is used to generate the corresponding elasticity-granting process architecture chart by performing the following operations on the control parameters at each point in the second control parameter sequence: The aforementioned operation is, The process involves performing a normalization operation on each of the multiple sub-parameters in the control parameter, mapping the sub-parameters to a predetermined value interval, and obtaining the normalized values ​​of the sub-parameters. This includes scaling the processing path in the POY yarn elasticity imparting process proportionally to the initialization image, and obtaining an elasticity imparting process architecture chart corresponding to the said time point. The apparatus according to claim 9, wherein the values ​​of pixel points other than the processing path in the elasticity imparting process architecture chart are set to default values, the points on the processing path corresponding to the subparameters in the control parameters are set to the normalized value of the subparameters, and the points on the processing path other than the subparameters in the control parameters are set to target values ​​different from the default values.

11. The aforementioned prediction subunit is used to perform the following operations: The aforementioned operation is, Based on a short-term feature extraction module constructed by the attention mechanism in the indicator prediction model, features are extracted within a first feature extraction range in the plurality of elasticity-granting process architecture charts to obtain short-term features, wherein the first feature extraction range includes elasticity-granting process architecture charts at n time points centered on the elasticity-granting process architecture chart at the target time point, where n is a positive integer, and the short-term feature extraction module uses the elasticity-granting process architecture chart at the target time point as a query feature, and the elasticity-granting process architecture charts at the n time points excluding the target time point as key features and value features, to obtain the short-term features, and Based on a long-term feature extraction module constructed by the attention mechanism in the indicator prediction model, features are extracted within a second feature extraction range in the plurality of elasticity-granting process architecture charts to obtain long-term features, wherein the second feature extraction range includes m elasticity-granting process architecture charts at various time points centered on the elasticity-granting process architecture chart at the target time point, where m is a positive integer greater than n, and the long-term feature extraction module divides the m elasticity-granting process architecture charts at various time points using a sliding window mechanism, determines pooling information within each sliding window, constructs key features and value features necessary for the attention mechanism based on the pooling information of the plurality of sliding windows, and obtains the long-term features based on the key features, value features, and query features. Using a fusion module, a fusion operation is performed on the long-term features and the short-term features to obtain the fused features. The apparatus according to claim 9, further comprising processing the fused features based on a prediction module in an indicator prediction model to obtain the indicator prediction result.

12. The aforementioned indicator prediction model further includes a supplementary feature extraction module, The aforementioned prediction subunit is further used to perform the following processing: The aforementioned process is, Based on the aforementioned supplementary feature extraction module, the following operations are performed to obtain supplementary features: This includes performing a fusion operation on the long-term features, short-term features, and supplementary features using a fusion module to obtain the fused features, The aforementioned operation is, The data is analyzed for the second control parameter sequence, and the cumulative difference of the parameters at each time point in the second control parameter sequence relative to the first time point in the second control parameter sequence is obtained to obtain a cumulative difference sequence. Based on the cumulative difference sequence, the time length relative to the first time point at which the cumulative difference is greater than a predetermined threshold is determined, and a time length sequence is obtained. The apparatus according to claim 11, further comprising extracting features from the second control parameter sequence, the cumulative difference sequence, and the time length sequence, and obtaining the supplementary features.

13. A training device for a control network of an elasticity imparting process, The control network includes a parameter prediction subnetwork, an indicator prediction subnetwork, and a data generation subnetwork. The training device is, An acquisition unit for acquiring a first training sample including predetermined control parameters at multiple time points in the elasticity imparting process, An input unit for inputting the first training sample into the parameter prediction subnetwork to obtain a parameter prediction value for at least one future time point, A processing unit for processing the parameter prediction values ​​based on the indicator prediction subnetwork and obtaining the predicted indicator, A training loss unit for determining the training loss based on the difference between the predicted indicator and the actual indicator, and the difference between the predicted parameter value and the actual parameter value, An adjustment unit for adjusting the parameter prediction subnetwork and the indicator prediction subnetwork based on the training loss, If the training convergence condition is met, a first generation unit for obtaining a parameter prediction model corresponding to the parameter prediction subnetwork and an index prediction model corresponding to the index prediction subnetwork, The system includes a second generation unit for training the data generation subnetwork based on the parameter prediction model and the index prediction model to obtain a data generation model, The data generation model generates desired control parameters for the elasticity imparting machine based on the index prediction results output from the index prediction model. The aforementioned indicator prediction results include at least one of the grade, pass rate, full winding rate, and dyeing uniformity of the wound yarn package, and are used in training the control network of the elasticity imparting process.

14. The aforementioned data generation subnetwork includes a training queue generator and a training queue discriminator, The second generation unit is used to perform multiple training cycles on the data generation subnetwork. Each of the aforementioned training sessions will involve performing the following operations: The aforementioned operation is, The true control parameter is input to the training-waiting discriminator, and a first discrimination result for the true control parameter is obtained by the training-waiting discriminator. The training wait generator is used to generate false control parameters, The false control parameters are input to the training-waiting classifier to obtain a second classification result. Adjusting the model parameters of the training-waiting classifier based on the first and second classification results, The apparatus according to claim 13, further comprising fixing the model parameters of the training waiting discriminator and training the training waiting generator.

15. It is an electronic device, At least one processor, Includes memory that is communicably connected to at least one processor, An electronic device having memory that stores instructions executable by at least one processor, the instructions being executed by at least one processor to cause at least one processor to perform the method according to any one of claims 1 to 5.

16. A non-temporary, computer-readable storage medium storing computer commands, wherein the computer commands are used to cause a computer to perform the method described in any one of claims 1 to 5.

Images