Control method for texturing process, training method for control network, and related apparatuses

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] The present disclosure relates to the technical field of data processing, and more particularly to a method for controlling an elasticizing process, a method for training a control network, and related devices. [Background technology]

[0002] In the elasticity imparting (false twisting) process, POY (Pre Oriented Yarn), abbreviated as POY raw yarn for the convenience of subsequent transportation and management, is processed into DTY (Draw Textured Yarn), and the DTY yarn is wound on a paper spool to form a DTY wound yarn package.

[0003] There are many controlled elements in the elasticizing process, and the parameter settings and management control of the corresponding elements have a significant impact on the quality of the final wound yarn package. Summary of the Invention

[0004] The present disclosure provides a method for controlling an elasticization process, a method for training a control network, and related apparatus to improve elasticization process flow.

[0005] According to one aspect of the present disclosure, there is provided a method for controlling an elasticity imparting process, the method comprising: sequentially collecting predetermined control parameters at a plurality of time points in an elasticization process flow to obtain a first sequence of control parameters; inputting the first sequence of control parameters into a parameter prediction model to predict predicted parameters for a plurality of future time points; selecting a target parameter for a target time point from the predicted parameters for the plurality of time points; constructing a second control parameter sequence including the target parameter based on the control parameters at a plurality of time points before and after the target time point; Processing the second control parameter sequence based on the indicator prediction model to obtain an indicator prediction result; If the index prediction result does not meet the desired value, inputting the index prediction result into a data generation model to obtain the desired control parameter for the target time point.

[0006] According to one aspect of the present disclosure, there is provided a method for training a control network of an elasticization process, the control network including a parameter prediction sub-network, an indicator prediction sub-network, and a data generation sub-network, the method comprising: obtaining a first training sample comprising predetermined control parameters at a plurality of time points of the elasticization process; inputting the first training sample into a parameter prediction sub-network to obtain parameter predictions for at least one time point in the future; processing the parameter predictions based on the indicator prediction sub-network to obtain a predicted indicator; determining a training loss based on the difference between the predicted indicators and the actual indicators and the difference between the parameter predictions and the actual parameter values; adjusting the parameter prediction sub-network and the indicator prediction sub-network based on the training loss; If the training convergence condition is satisfied, obtaining a parameter prediction model corresponding to the parameter prediction sub-network and an indicator prediction model corresponding to the indicator prediction sub-network; training a data generation sub-network based on the parameter prediction model and the indicator prediction model to obtain a data generation model; The data generation model is used to generate desired control parameters for the elasticity applying machine based on the index prediction results output from the index prediction model.

[0007] According to another aspect of the present disclosure, there is provided a control device for an elasticizing process, the control device comprising: a collection unit for sequentially collecting predetermined control parameters at multiple time points in the elasticizing process to obtain a first sequence of control parameters; a parameter prediction unit for inputting the first control parameter sequence into a parameter prediction model to predict predicted parameters for a plurality of future time points; a selection unit for selecting a target parameter for a target time point from the predicted parameters for the plurality of time points; a construction unit for constructing a second control parameter sequence including the target parameter based on the control parameters at a plurality of time points before and after the target time point; an index prediction unit for processing the second control parameter sequence based on the index prediction model to obtain an index prediction result; and a data generating unit for inputting the index prediction result into a data generating model to obtain a desired control parameter for the target time point if the index prediction result does not satisfy the desired value.

[0008] According to one aspect of the present disclosure, there is provided a training device for a control network of an elasticization process, the control network including a parameter prediction sub-network, an indicator prediction sub-network, and a data generation sub-network, the device comprising: an acquisition unit for acquiring a first training sample including predetermined control parameters at multiple time points of the elasticization process; an input unit for inputting the first training sample into the parameter prediction sub-network to obtain parameter prediction values for at least one time point in the future; a processing unit for processing the parameter prediction values based on the indicator prediction sub-network to obtain a predicted indicator; a training loss unit for determining a training loss based on the difference between the predicted indicators and the actual indicators and the difference between the parameter predicted values and the actual parameter values; an adjustment unit for adjusting the parameter prediction sub-network and the indicator prediction sub-network based on the training loss; a first generation unit for obtaining a parameter prediction model corresponding to the parameter prediction sub-network and an indicator prediction model corresponding to the indicator prediction sub-network when the training convergence condition is satisfied; a second generating unit for training a data generating sub-network based on the parameter prediction model and the indicator prediction model to obtain a data generating model; The data generation model is used to generate desired control parameters for the elasticity applying machine based on the index prediction results output from the index prediction model.

[0009] According to another aspect of the present disclosure, there is provided an electronic device, the electronic device comprising: At least one processor; a memory communicatively connected to the at least one processor; The memory stores instructions executable by the at least one processor, the instructions being 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, there is provided a non-transitory computer-readable storage medium having stored thereon computer instructions for causing the computer to perform a method according to any embodiment of the present disclosure.

[0011] According to another aspect of the present disclosure, there is provided a computer program product including a computer program that, when executed by a processor, implements a method according to any embodiment of the present disclosure.

[0012] In an embodiment of the present disclosure, predetermined control parameters are collected, and a parameter prediction model is used to predict future parameters, and a target parameter is selected to construct a second control parameter sequence. The second control parameter sequence is processed using an indicator prediction model to obtain an indicator prediction result, and if the desired value is not met, a data generation model is used to obtain the desired control parameter, thereby adjusting the parameters in the elasticization process.

[0013] It should be understood that the contents of the Summary of the Invention section do not limit the key points or important features of the embodiments of the present disclosure, nor do they limit the scope of the present disclosure. Other features of the present disclosure will be readily understood from the following specification.

[0014] The drawings are for a better understanding of the present technical solution, and are not intended to limit the present disclosure. [Brief explanation of the drawings]

[0015] [Figure 1] FIG. 1 is a schematic diagram of an elasticity applicator according to one embodiment of the present disclosure. [Figure 2] FIG. 2 is a flow chart of a method for controlling an elasticization process according to one embodiment of the present disclosure. [Figure 3] FIG. 3 is a flowchart of a method for training a control network of an elasticization process according to one embodiment of the present disclosure. [Figure 4] FIG. 4 is a structural schematic diagram of a training parameter prediction sub-network according to one embodiment of the present disclosure. [Figure 5] FIG. 5 is a flowchart of obtaining an elasticity granting process architecture chart according to one embodiment of the present disclosure. [Figure 6] FIG. 6 is a schematic diagram of an elasticity imparting process architecture chart obtained by a second control parameter sequence according to an embodiment of the present disclosure. [Figure 7] FIG. 7 is a flowchart for providing a feature module according to one embodiment of the present disclosure. [Figure 8] FIG. 8 is a structural schematic diagram of a control device for an elasticity imparting process according to an embodiment of the present disclosure. [Figure 9] FIG. 9 is a structural schematic diagram of a control network training device for elasticity application process according to an embodiment of the present disclosure. [Figure 10] FIG. 10 is a block diagram of electronics for implementing the elasticization process control method / control network training method of an embodiment of the present disclosure. DETAILED DESCRIPTION OF THE INVENTION

[0016]

[0023] Exemplary embodiments of the present disclosure will now be described with reference to the drawings. While various details of the embodiments of the present disclosure are described for ease of understanding, it should be understood that they are merely illustrative. Therefore, those skilled in the art should recognize 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 conciseness, descriptions of well-known functions and structures will be omitted in the following description.

[0017] It should be noted that terms such as "first," "second," and the like in the specification, claims, and drawings of this application are intended to distinguish between similar objects and are not intended to describe a particular order or sequence. It should be understood that such terms may be interchanged under appropriate circumstances, such that the embodiments of the application described herein may be performed in an order other than that shown or described herein. The embodiments described in the following illustrative examples are not intended to represent all embodiments consistent with this application. Rather, they are merely examples of devices and methods consistent with certain aspects of this application, as detailed in the appended claims.

[0018] A schematic diagram of an exemplary elasticity imparting machine is as shown in FIG. 1, and the main components of the elasticity imparting 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 twisting member 106, a nozzle 107, a second roller 108, a second heating box 109, a third roller 110, a yarn breakage detecting device 111, and a winding member 112.

[0019] The first roller 103, the second roller 108, and the third roller 110 are for ensuring that the yarn is processed along a predetermined path. In the elasticizing process flow, the speeds of the first roller 103, the second roller 108, and the third roller 110 are coordinated with each other to prevent the yarn from being torn or piled up.

[0020] Depending on the needs of the product form, the nozzle 107 may be used to process the processed yarn through a false twisting member into a network-shaped yarn to achieve the desired feel and form for different yarns.

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

[0022] If the yarn breakage detector 111 detects a yarn breakage, it triggers the yarn cutter 102 to cut the yarn to avoid accumulation in the subsequent flow of POY yarn.

[0023] The elasticization 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 between the speed of the second roller and 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 the second roller. 4) The false twist ratio, which is the ratio between the number of rotations of the disks in the false twist member and the operating speed. 5) Oil wheel rotation speed, 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 used to supply oil to the yarn. 6) Full winding time, which is the time required for the winding element to complete winding of one full package of yarn. 7) The upper heating box temperature H1 is the temperature of the modified heating box (i.e., the first heating box). 8) Lower heating box temperature H2, which is the temperature of the shape-stabilizing heating box (i.e., the second heating box). 9) Air pressure value, which is the air pressure value required for the nozzle to hit the network.

[0024] Of course, the above parameters are merely examples, and the corresponding key parameters can be determined according to the specific circumstances during implementation.

[0025] The impact of the elasticization process on the quality of DTY wound yarn packages is a complex, systematic issue. Inappropriate elasticization process parameters can lead to defects in the final product. Therefore, a method is needed to optimize the elasticization process and improve product quality and production efficiency in order to timely detect and address problems during production.

[0026] In view of this, an embodiment of the present disclosure provides a method for controlling the elasticity imparting process to solve the above problem.

[0027] As shown in FIG. 2, the method for controlling the elasticity imparting process according to an embodiment of the present disclosure includes:

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

[0029] In the embodiment of the present disclosure, the predetermined control parameters include the 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, air pressure value, etc.

[0030] S202: The first control parameter sequence is input to a parameter prediction model to predict predicted parameters at multiple time points in the future.

[0031] S203: A target parameter for a target time point is selected from the predicted parameters for the plurality of time points.

[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 index prediction model to obtain an index prediction result.

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

[0035] In an embodiment of the present disclosure, predetermined control parameters for multiple time points in the elasticization process flow are collected and input into a parameter prediction model as a first control parameter sequence to predict predicted parameters for multiple future time points. Then, a target parameter for the target time point is selected from these predicted parameters. A second control parameter sequence including the target parameter is constructed based on the control parameters for multiple time points before and after the target time point. By processing the second control parameter sequence using the index prediction model, an index prediction result for the future target time point can be predicted. If the index prediction result does not meet the desired value, possible future parameter problems can be discovered in a timely manner, and the index prediction result can be input into a data generation model to obtain the desired control parameters for the future target time point. Predicting and adjusting the control parameters contributes to optimizing the elasticization process flow and ensuring that the quality of the final product meets expectations as much as possible.

[0036] For better understanding, we will explain it below from two aspects: training the model and using the model.

[0037] 1. Model training

[0038] In the embodiment of the present disclosure, the parameter prediction model, the indicator prediction model, and the data generation model are obtained by training a control network, which includes a parameter prediction sub-network (for generating the corresponding parameter prediction model after being trained), an indicator prediction sub-network (for generating the corresponding indicator prediction model after being trained), and a data generation sub-network (for generating the corresponding data generation model after being trained).

[0039] Based on this, an embodiment of the present disclosure provides a method for training a control network of an elasticization process, which may include the following steps, as shown in FIG.

[0040] S301: Obtain a first training sample including predetermined control parameters at multiple time points of the elasticization process.

[0041] For example, data collection may be performed on key parameters during the actual elasticization process, and multiple predetermined control parameters in the elasticization process may be obtained at the same time point, where these predetermined control parameters are key parameters in the elasticization process.

[0042] A first training sample corresponding to time t can be constructed around time t, for example, by selecting predetermined control parameters corresponding to L time points to construct the first training sample, for example, by selecting predetermined control parameters corresponding to time points t2, t-1, t, t+1, and t+2 to construct the first training sample.

[0043] This allows for the construction of a corresponding first training sample for each sampling time point.

[0044] S302: Input a first training sample into a parameter prediction sub-network to obtain parameter predictions for at least one time point in the future.

[0045] In the embodiment of the present disclosure, the Koopman time series model is selected as the parameter prediction sub-network. The specific training steps are as follows:

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

[0047] For example, if the length of the training sample is T, a Fourier transform is performed on each sequence to obtain the amplitude value corresponding to each frequency in the spectrum set S = {0, 1, ..., {T / 2}}. For each frequency, the amplitude value corresponding to that frequency is averaged to obtain the magnitude of the dominant frequency in the entire sample (the frequency with the highest average amplitude value can be determined as the dominant frequency). The set g obtained by selecting the frequencies with the highest average amplitude value and a proportion α is α the time-invariant component X inv Then, the remaining part is the time-varying component X VAR The specific steps are as shown in the following formula (1).

[0048]

number

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

[0050] S3022: For the time-invariant component, use a 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 4a, the time-invariant components are input to the time-invariant Koopman predictor, and the time-invariant components are converted to R by the Encoder. D , that is, as shown in equation (2) below.

[0052]

number

[0053] Next, in the latent space, the size is R D×D A learnable matrix K inv Use the conversion, i.e.

[0054]

number

[0055] Finally, the decoder generates R H×C , i.e.,

[0056]

number

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

[0058] S3023: For the time-varying component, a time-varying Koopman predictor is used 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. To improve the processing efficiency, the input time-varying component is divided into multiple segments of length S (the figure shows an example in which it is divided into three segments), and each segment is X j Labeled as X j is from any element in all segments and can be expressed as the following equation (5):

[0060]

number

[0061] In equation (5), X j indicates the time-varying component of each divided segment.

[0062] We use a pair of encoders / decoders to learn the transformation of high-dimensional linear systems, i.e.,

[0063]

number

[0064] In equation (5), Z j denotes the corresponding features obtained by the corresponding encoder of the corresponding segment, JPEG2025115387000008.jpg86 is Here are the new variables obtained by decoding JPEG2025115387000009.jpg86.

[0065] According to the obtained corresponding features, a data-driven EDMD (mode decomposition algorithm) algorithm is used to obtain the time-varying operator, namely, as shown in Equation (7) below:

[0066]

number

[0067] Based on the obtained operator, the fitting result can be calculated for the previously observed dynamic system, namely, as shown in the following equation (8):

[0068]

number

[0069] Furthermore, by extrapolating, we can predict the future results Z T / S+t That is, as shown in the following equation (9).

[0070]

number

[0071] Finally, the fitting output and sequence prediction of the module can be obtained by combining these segments anew, i.e., as shown in Equation (10) below.

[0072]

number

[0073] In (10), JPEG2025115387000014.jpg912 is the input data for the next KoopaBlock layer, and Y var is the predicted value of the second parameter at a future time point.

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

[0075] S303: Process the parameter prediction value based on the indicator prediction sub-network to obtain a predicted indicator, where the implementation can include the following steps:

[0076] S3031: If the parameter prediction value is a third parameter sequence constructed using parameters for multiple time points, a specified parameter for a specified time point is selected from the third parameter sequence, and based on the obtained specified parameter, control parameter samples for multiple time points based on the time point of the specified parameter are constructed to obtain a fourth control parameter sequence.

[0077] S3032: Process the fourth control parameter sequence based on the indicator prediction sub-network to obtain a predicted indicator.

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

[0079] Here, the following operations (including step A1 to step A2) are performed for each control parameter at each time point in the obtained fourth control parameter sequence, thereby generating a corresponding elasticity-imparting process architecture chart.

[0080] Step A1: A normalization operation is performed on each of a plurality of subparameters in the control parameter, thereby mapping the subparameters to a predetermined value interval and obtaining normalized values of the subparameters.

[0081] Step A2: The processing path in the elasticization process of the POY yarn is scaled to the initialization image in equal proportions (equal scale) to obtain the elasticization process architecture chart corresponding to the time point.

[0082] Here, the values of pixel points other than the processing path in the elasticity-imparting process architecture chart take default values, points corresponding to subparameters in the control parameters on the processing path are set to the normalized values of the subparameters, and 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-mentioned step S3032 can be implemented 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 sub-network, extract features within a third feature extraction range in the multiple elasticity-granted process architecture charts to obtain short-term features, where the third feature extraction range includes elasticity-granted process architecture charts for n time points centered on the elasticity-granted process architecture chart for the target time point, where n is a positive integer, and the short-term feature extraction module uses the elasticity-granted process architecture chart for the target time point as a query feature and the elasticity-granted process architecture charts for the 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 sub-network, extract features within a fourth feature extraction range in the multiple elasticity-granting process architecture charts to obtain long-term features, where the fourth feature extraction range includes elasticity-granting process architecture charts of m time points centered on the elasticity-granting process architecture chart of the target time point, where m is a positive integer greater than n, and the long-term feature extraction module divides the elasticity-granting process architecture charts of the m time points using a sliding window mechanism to determine pooling information within each sliding window, and constructs key features and value features required for the attention mechanism based on the pooling information of the multiple sliding windows, and obtains long-term features based on the key features, value features, and query features.

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

[0087] Step B4: Process the fused features based on the prediction module of the indicator prediction sub-network to obtain a predicted indicator.

[0088] In order to make the predicted index results obtained by the index prediction sub-network processing the parameter prediction values more accurate, the index prediction sub-network further includes a supplementary feature extraction module, which can perform the following operations to obtain supplementary features based on the obtained parameter prediction values:

[0089] Step C1: Perform data analysis on the obtained fourth control parameter sequence, and obtain a cumulative difference sequence by obtaining a cumulative difference sequence of parameters at each time point in the fourth control parameter sequence relative to a first time point in the fourth control parameter sequence, where the parameter at the first time point is the first control parameter in the fourth control parameter sequence.

[0090] Step C2: Determine the time length of the time point in the fourth control parameter sequence at which the accumulated difference is greater than the predetermined threshold value relative to the first time point, based on the accumulated difference sequence, to obtain a time length sequence.

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

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

[0093] S304: Determine a training loss based on the difference between the predicted index and the actual index, and the difference between the parameter predicted values and the actual parameter values.

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

[0095] S306: If the training convergence condition is met, obtain a parameter prediction model corresponding to the parameter prediction sub-network and an indicator prediction model corresponding to the indicator prediction sub-network.

[0096] In an embodiment of the present disclosure, the mean square error can be used as a loss function to account for the difference between the predicted index and the actual index, and the difference between the parameter predicted value and the actual parameter value, thereby adjusting the parameter prediction sub-network and the index prediction sub-network. Finally, the model is trained through iterations, and the performance of the model is evaluated using a validation set or a test set. If the requirements are met, the optimized parameter prediction sub-network is determined as the parameter prediction model, and the optimized index prediction sub-network is determined as the index prediction model.

[0097] S307: Train a data generation sub-network based on the parameter prediction model and the index prediction model to obtain a data generation model, which is for generating desired control parameters of the elasticity applying machine based on the index prediction result output from the index prediction model, where the generation sub-network includes a training wait generator and a training wait classifier, and trains the data generation sub-network based on the parameter prediction model and the index prediction model to obtain a data generation model.

[0098] To ensure the accuracy of the resulting data generation model, the data generation sub-network needs to be trained multiple times, and each training session performs the following operations:

[0099] Step D1: Input the true control parameters into a training classifier to obtain a first classification result for the true control parameters by the classifier.

[0100] Step D2: Generate fake control parameters using a trained generator.

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

[0102] Step D4: Adjust the model parameters of the classifiers waiting to be trained based on the first and second discrimination results.

[0103] For example, a waiting-to-train generator is used to generate false control parameters, and the generated false control parameters are combined with true control parameters to form a true / false dataset. A waiting-to-train discriminator is trained based on the constructed true / false dataset without changing the generator parameters. For example, a random training sample is given and the waiting-to-train discriminator is trained so that it can determine whether the sample is a true control parameter or a false sample generated by the waiting-to-train generator. The true control parameters are input into the waiting-to-train discriminator, and the obtained discrimination result is taken as the true result. The obtained false control parameters are input into the waiting-to-train discriminator, and the obtained discrimination result is taken as the false result. The training of the waiting-to-train discriminator is completed by referring to the true result and the false result.

[0104] Step D5: Fix the model parameters of the training-waiting classifier and train the training-waiting generator.

[0105] After the training of the training-awaiting classifier is completed, a training-awaiting generator is connected in series to the trained training-awaiting classifier, and the training-awaiting generator is trained while the parameters of the training-awaiting classifier are fixed. The training-awaiting generator is given the predicted index output from the index prediction model to obtain the control parameters generated by the training-awaiting generator, and the generated control parameters are input to the training-awaiting classifier for judgment. A loss value is generated as a target for processing by a loss function based on the recognition result of the training-awaiting classifier, and the parameters of the training-awaiting generator are updated based on the loss value.

[0106] After completing the training of the trained generator, the trained classifier is trained again while keeping the parameters of the trained generator fixed. The above process is repeated until the trained generator can generate sufficiently realistic data and the trained classifier cannot accurately distinguish between real and false data. This results in a data generation model.

[0107] In an embodiment of the present disclosure, the training generator and the training classifier process the predicted parameters generated by the parameter prediction model and the indicator prediction results generated by the indicator prediction model to obtain a data generation model, and provide efficient optimization efficiency for parameters in the elasticity assignment process flow.

[0108] In an embodiment of the present disclosure, a parameter prediction subnetwork is trained using a first training sample to obtain parameter prediction values. An indicator prediction subnetwork is trained based on the obtained parameter prediction values to obtain indicator parameters. A training loss is determined to adjust the parameter prediction subnetwork and the indicator prediction subnetwork to obtain a parameter prediction model and an indicator prediction model. Based on this, a data generation subnetwork can be trained to obtain a data generation model. Training the subnetwork to obtain a final prediction model improves the optimization efficiency for parameters in the elasticity assignment flow.

[0109] 2. Use of the Model

[0110] The POY yarn passes through a yarn processing path and completes the elasticization process to obtain a desired DTY yarn. The entire yarn path is designed based on the elasticization process of the elasticization machine. In addition to the processing of corresponding key parts in the entire yarn processing, some mechanical characteristics of the elasticization machine and the processing process of the POY yarn in the yarn path also affect the final yarn quality. To uncover the impact of potential hidden features on DTY yarn quality, in an embodiment of the present disclosure, processing the second control parameter sequence based on the index prediction model and obtaining the index prediction result can be implemented by generating corresponding elasticization process architecture charts based on the control parameters at each time point of the second control parameter sequence to obtain multiple elasticization process architecture charts, and then inputting the multiple elasticization process architecture charts into the index prediction model to obtain the index prediction result.

[0111] In the embodiment of the present disclosure, the elasticizing process architecture chart can embody the yarn processing path of the elasticizing machine and the key parameters of the key parts, so that by linking the predicted control data with the elasticizing process path of the elasticizing machine and converting it into a diagram format, the potential features and relationships can be mined, and the quality at a future time point can be easily predicted.

[0112] In implementation, the generation of the elasticity-imparting process architecture chart is similar to that during model training, and as shown in Figure 5, for the control parameters at each time point in the second control parameter sequence, the following operations as shown in Figure 5 are respectively performed to generate the corresponding elasticity-imparting process architecture chart.

[0113] S501: A normalization operation is performed on each of a plurality of subparameters in the control parameter, and the subparameters are mapped to predetermined value intervals to obtain normalized values of the subparameters.

[0114] For example, for each subparameter, the maximum and minimum values are calculated to determine the range of data for the subparameter.

[0115] The normalization operation for the 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] In order to reduce the amount of calculation in the subsequent training process, the new calculated values are mapped to a predetermined value interval, which in this embodiment 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 elasticization process of the POY yarn is scaled proportionally to the initialization image to obtain an elasticization process architecture chart corresponding to the time point, wherein the values of pixel points other than the processing path in the elasticization process architecture chart take default values, points corresponding to sub-parameters in the control parameters on the processing path are set to normalized values of the sub-parameters, and points other than the sub-parameters in the control parameters on the processing path are set to target values different from the default values.

[0120] In the embodiment of the present disclosure, the default value is set to 255, which is the value of pixel points other than the machining path in the initialization image, the target value is set to 0, which is the value of points other than the subparameters in the control parameters on the machining path, and the remaining values are the values of points corresponding to the subparameters in the control parameters on the machining path, so as to obtain an elasticity imparting process architecture chart as shown in Fig. 6. It may be understood that the entire elasticity imparting process flow and corresponding components are scaled to the initial image, and the values of important parts in the diagram can represent the process parameters of the parts.

[0121] In an embodiment of the present disclosure, a normalization operation is performed on multiple subparameters in the second control parameter sequence, mapping them to a predetermined value interval to obtain normalized values. The processing path of the POY yarn elasticization process is then proportionally scaled onto the initialization image. Pixel points outside the processing path in the initialization image take default values, while points on the processing path corresponding to subparameters in the control parameters are set to the normalized values of those subparameters, and points other than the subparameters are set to target values different from the default values. In this way, an elasticization process architecture chart is obtained. Converting real-time control parameters into a graph representation provides a good data basis for subsequent indicator prediction models to uncover potential correlations and features.

[0122] In some embodiments, in order to better mine features and improve the accuracy of indicator prediction, multiple elastic process architecture charts are input into the indicator prediction model to obtain indicator prediction results, which can be specifically implemented as shown in FIG. 7.

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

[0124] For example, given a series of elastic process architecture charts, each chart representing the process state at a particular point in time, in embodiments of the present 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, the first feature extraction range, i.e., the elasticity-imparting process architecture chart for n time points centered on the target time point, is determined. If five time points are selected as the first feature extraction range, two time points are selected before and after the current time point (target time point).

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

[0127] S702: Based on a long-term feature extraction module constructed by the attention mechanism in the indicator prediction model, extract features within a second feature extraction range in multiple elastically-enhanced process architecture charts to obtain long-term features, where the second feature extraction range includes elastically-enhanced process architecture charts for m time points centered on the elastically-enhanced process architecture chart at the target time point, where m is a positive integer greater than n, and the long-term feature extraction module uses a sliding window mechanism to divide the elastically-enhanced process architecture charts for m time points to determine pooling information within each sliding window, construct key features and value features required for the attention mechanism based on the pooling information of the multiple sliding windows, and obtain long-term features based on the key features, value features, and query features.

[0128] For example, there may be an elastic process architecture chart sequence containing 100 time points, each of which represents a different process state.

[0129] Therefore, first, the second feature extraction range, i.e., the elasticity-imparting process architecture chart for m time points centered on the target time point, is determined. If 20 time points are selected as the second feature extraction range, 10 time points before and after the current time point (target time point) are selected.

[0130] Next, a long-term feature extraction module is used to extract features. The module uses a sliding window mechanism to divide the elasticity-imparted process architecture chart into 20 time points and determine pooling information within each sliding window. For example, a window size can be set to 5, and then the window can be slid across these 20 time points. Also, assuming that the windows can overlap or not, pooling information for at least four windows can be obtained. In each window, the model constructs key features and value features based on an attention mechanism.

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

[0132] S704: Process the fusion feature based on the prediction module to obtain an index prediction result.

[0133] In an embodiment of the present disclosure, a short-term feature extraction module extracts short-term features from multiple elasticity application process architecture charts, a long-term feature extraction module extracts long-term features from multiple elasticity application processes, and the obtained short-term and long-term features are fused as input parameters for a prediction module to obtain indicator prediction results. By extracting and fusing long-term and long-term features, when processing data parameters with different time scales, data characteristics can be more comprehensively captured, thereby improving the prediction performance and effectiveness of the elasticity application process flow of the prediction model.

[0134] Furthermore, in some embodiments, in order to make the index prediction results obtained by the prediction model more accurate, the index prediction model further includes a supplementary feature extraction module, which performs the following operations based on the supplementary feature extraction module to obtain supplementary features:

[0135] Specifically, the supplementary feature extraction module performs data analysis on the second control parameter sequence, obtains cumulative differences of parameters at each time point in the second control parameter sequence relative to a first time point in the second control parameter sequence to obtain a cumulative difference sequence, determines, based on the cumulative difference sequence, the time lengths relative to the first time point at which the cumulative difference is greater than a predetermined threshold, obtains a time length sequence, and performs feature extraction on the second control parameter sequence, the cumulative difference sequence, and the time length sequence to obtain supplementary features.

[0136] For example, there may be one second control parameter sequence that includes 10 time points, each time point representing a different control parameter value.

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

[0138] Next, the data analysis is performed on the second control parameter sequence to 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 is 2. By this analogy, a new cumulative difference sequence is obtained.

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

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

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

[0142] In the embodiment of the present disclosure, by extracting supplementary features, more information about the second control parameter sequence can be obtained, helping the model better understand and predict future elasticity-imparting process architecture charts. The addition of the extraction module further optimizes the predictive model, making the results of the predictive model more accurate.

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

[0144] In the embodiments of the present disclosure, key indicators such as the grade, pass rate, full winding rate, and dye uniformity of the wound yarn package can be predicted, allowing problems to be detected in a timely manner and addressed and improved, thereby reducing the reject rate and lowering production costs.

[0145] Based on the same technical concept, an embodiment of the present disclosure provides a control device for the elasticity imparting process, as shown in FIG. 8: a collection unit 801 for sequentially collecting predetermined control parameters at multiple time points in an elasticity imparting process flow to obtain a first control parameter sequence; a parameter prediction unit 802 for inputting the first control parameter sequence into a parameter prediction model to predict predicted parameters for a plurality of future time points; a selection unit 803 for selecting a target parameter at a target time point from the predicted parameters at the multiple time points; a construction unit 804 for constructing a second control parameter sequence including the target parameter based on the control parameters at a plurality of time points before and after the target time point; an index prediction unit 805 for processing the second control parameter sequence based on the index prediction model to obtain an index prediction result; and a data generation unit 806 for inputting the index prediction result into a data generation model to obtain desired control parameters for the target time point if the index prediction result does not satisfy the desired value.

[0146] In some embodiments, the indicator prediction unit comprises: a generating subunit for generating a corresponding elasticity-imparted process architecture chart according to the control parameters at each time point of the second control parameter sequence, thereby obtaining a plurality of elasticity-imparted process architecture charts; and a prediction subunit for inputting the plurality of elasticity-imparting process architecture charts into an index prediction model and obtaining an index prediction result.

[0147] In some embodiments, the product subunit specifically comprises: For each control parameter at each time point in the second control parameter sequence, the following operations are performed to generate a corresponding elasticity-imparting process architecture chart: The operation is: performing a normalization operation on each of a plurality of sub-parameters of the control parameter, mapping the sub-parameters to a predetermined value interval, and obtaining normalized values of the sub-parameters; and proportionally scaling the processing path in the elasticizing process of the POY yarn onto the initialization image to obtain an elasticizing 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 default values, points corresponding to subparameters in the control parameters on the processing path are set to the normalized values of the subparameters, and 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 specifically used to perform the following operations: The operation is: extracting features within a first feature extraction range in the multiple elasticity-granted process architecture charts based on a short-term feature extraction module constructed by an attention mechanism in the indicator prediction model to obtain short-term features, wherein the first feature extraction range includes elasticity-granted process architecture charts at n time points centered on the elasticity-granted process architecture chart at the target time point, where n is a positive integer, and the short-term feature extraction module uses the elasticity-granted process architecture chart at the target time point as a query feature and the elasticity-granted process architecture charts at the n time points excluding the target time point as key features and value features to obtain short-term features; extracting features within a second feature extraction range in the multiple elasticity-granting process architecture charts based on a long-term feature extraction module constructed by the attention mechanism in the indicator prediction model to obtain long-term features, where the second feature extraction range includes elasticity-granting process architecture charts of m time points centered on the elasticity-granting process architecture chart of the target time point, where m is a positive integer greater than n, and the long-term feature extraction module uses a sliding window mechanism to divide the elasticity-granting process architecture charts of the m time points to determine pooling information in each sliding window, and construct key features and value features required for the attention mechanism based on the pooling information of the multiple sliding windows, and obtain long-term features based on the key features, value features, and query features; performing a fusion operation on the long-term features and the short-term features using a fusion module to obtain fused features; and processing the fused feature based on a prediction module in the index prediction model to obtain an index prediction result.

[0149] In some embodiments, the index prediction model further comprises a supplementary feature extraction module, and the prediction subunit is further used to perform the following processing: The process is as follows: Obtaining supplementary features by performing the following operations based on a supplementary feature extraction module; and performing a fusion operation on the long-term features, the short-term features, and the supplemental features using a fusion module to obtain fused features; The operation is: performing an analysis of the data on the second control parameter sequence to obtain a cumulative difference of the parameter for each time point in the second control parameter sequence relative to the first time point in the second control parameter sequence to obtain a cumulative difference sequence; determining, based on the cumulative difference sequence, a time length relative to the first time point at which the cumulative difference is greater than a predetermined threshold, to obtain a time length sequence; and performing feature extraction on the second control parameter sequence, the cumulative difference sequence, and the time length sequence to obtain supplemental features.

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

[0151] An embodiment of the present disclosure provides a training device for a control network of an elasticity application process, the control network including a parameter prediction sub-network, an indicator prediction sub-network and a data generation sub-network, as shown in FIG. 9 : an obtaining unit 901 for obtaining a first training sample including predetermined control parameters at multiple time points of the elasticization process; an input unit 902 for inputting the first training sample into the parameter prediction sub-network to obtain parameter predictions for at least one time point in the future; a processing unit 903 for processing the parameter prediction values based on the indicator prediction sub-network to obtain a predicted indicator; a training loss unit 904 for determining a training loss based on the difference between the predicted indicators and the actual indicators and the difference between the parameter predicted values and the actual parameter values; a training loss unit 904 for adjusting the parameter prediction sub-network and the indicator prediction sub-network based on the training loss; a first generating unit 906 for obtaining a parameter prediction model corresponding to the parameter prediction sub-network and an indicator prediction model corresponding to the indicator prediction sub-network if the training convergence condition is met; a second generating unit 907 for training a data generating sub-network based on the parameter prediction model and the indicator prediction model to obtain a data generating model; The data generation model is used to generate desired control parameters for the elasticity applying machine based on the index prediction results output from the index prediction model.

[0152] In some embodiments, the data-generating sub-network includes a training-waiting generator and a training-waiting discriminator, and the second generating unit is specifically used to perform multiple training rounds on the data-generating network, and each training round performs the following operations: The operation is inputting the true control parameters into a training classifier to obtain a first classification result for the true control parameters by the classifier; generating fake control parameters using a trained generator; inputting the false control parameters into the trained classifier to obtain a second classification result; adjusting model parameters of the training classifier based on the first discrimination result and the second discrimination result; Fixing model parameters of the trained classifier and training the trained generator.

[0153] For specific functions and exemplary descriptions of each module, sub-module / unit of the apparatus according to the embodiments of the present disclosure, please refer to the relevant descriptions of the corresponding steps in the above method embodiments, and the description will be omitted here.

[0154] In the technical solution disclosed herein, the acquisition, storage, and application of personal information of relevant users shall comply with the provisions of relevant laws and regulations and shall not violate public order and morals.

[0155] FIG. 10 is a block diagram of an electronic device according to an embodiment of the present disclosure. As shown in FIG. 10, the electronic device includes a memory 1010 and a processor 1020. The memory 1010 stores a computer program executable 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 described in the above embodiment. The electronic device may further include a communication interface 1030 for communicating with an external device and exchanging data.

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

[0157] Optionally, as a specific implementation, when the memory 1010, the processor 1020, and the communication interface 1030 are integrated into one chip, the memory 1010, the processor 1020, and the communication interface 1030 can communicate with each other via an internal interface.

[0158] It should be understood that the processor may be a Central Processing Unit (CPU), other general-purpose processors, Digital Signal Processing (DSP), Application Specific Integrated Circuits (ASIC), Field Programmable Gate Arrays (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 conventional processor, etc. The processor may also be a processor capable of supporting an Advanced Reduced Instruction Set Machine (ARM) architecture.

[0159] Additionally, optionally, the memory may include read-only memory, random access memory, or non-volatile random access memory. The memory may be volatile or non-volatile memory, or may include both volatile and non-volatile memory. Non-volatile memory may include read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) used as an external cache. The above description is illustrative only 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 Rate 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 above embodiments, all or part of the above may be implemented by software, hardware, firmware, or any combination thereof. When implemented by software, all or part of the above 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 by a computer, they generate all or part of the flows or functions described in the embodiments of the present disclosure. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in 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 wire (e.g., coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (e.g., infrared, Bluetooth, microwave, etc.). The computer-readable storage medium may be any available medium accessible by a computer, or may be a data storage device such as a server, data center, or the like that includes one or more available media. The available medium may be a magnetic medium (e.g., a floppy disk, a hard disk, or a magnetic tape), an optical medium (e.g., a Digital Versatile Disk (DVD)), a semiconductor medium (e.g., a Solid State Disk (SSD)), or the like. Note that the computer-readable storage medium according to the present disclosure may be a non-volatile storage medium, in other words, a non-transitory storage medium.

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

[0162] In describing embodiments of the present disclosure, the terms "one embodiment," "some embodiments," "examples," "particular examples," or "some examples" 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. Furthermore, the described specific features, structures, materials, or characteristics can be combined in any suitable manner in any one or more embodiments or examples. Furthermore, unless mutually inconsistent, a person skilled in the art can combine features from different embodiments or examples described herein.

[0163] In the description of the embodiments of the present disclosure, unless otherwise specified, " / " means "or", for example, "A / B" can represent "A" or "B". The description of "and / or" in this specification is merely a relation describing related objects, and means that there may be three relations, for example, "A and / or B" can indicate three situations: "A" exists alone, "A" and "B" exist simultaneously, and "B" exists alone.

[0164] In describing the embodiments of the present disclosure, the terms "first" and "second" are for distinguishing purposes and should not be understood to indicate or imply relative importance or the number of the indicated components. Thus, a feature qualified with "first" or "second" can explicitly or implicitly include one or more of the feature. In describing the embodiments of the present disclosure, unless otherwise specified, "plurality" means two or more.

[0165] The above are merely illustrative examples 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 should be included within the scope of the claims of the present disclosure.

Claims

1. A method for controlling an elasticity imparting process, comprising: Sequentially collecting predetermined control parameters at a plurality of time points in an elasticization process flow to obtain a first sequence of control parameters; inputting the first sequence of control parameters into a parameter prediction model to predict predicted parameters for a plurality of future time points; selecting a target parameter for a target time point from the predicted parameters for the plurality of time points; constructing a second control parameter sequence including the target parameter based on control parameters at a plurality of time points before and after the target time point; processing the second control parameter sequence based on an index prediction model to obtain an index prediction result; If the index prediction result does not satisfy a desired value, inputting the index prediction result into a data generation model to obtain a desired control parameter for the target time point.

2. processing the second control parameter sequence based on the index prediction model to obtain an index prediction result, generating a corresponding elasticity-imparted process architecture chart according to the control parameters of each time point of the second control parameter sequence, thereby obtaining a plurality of elasticity-imparted process architecture charts; and inputting the plurality of elasticity application process architecture charts into the indicator prediction model to obtain the indicator prediction result.

3. generating a corresponding elasticity-imparted process architecture chart based on the control parameters of each time point of the second control parameter sequence to obtain a plurality of elasticity-imparted process architecture charts; For each control parameter at each time point in the second control parameter sequence, perform the following operations to generate a corresponding elasticity-imparting process architecture chart: The operation is performing a normalization operation on each of a plurality of sub-parameters of the control parameter, mapping the sub-parameters to a predetermined value interval, and obtaining normalized values of the sub-parameters; and proportionally scaling the processing path in the elasticizing process of the POY yarn onto the initialization image to obtain an elasticizing process architecture chart corresponding to said time point; 3. The method according to claim 2, wherein values of pixel points other than the processing path in the elasticity-imparting process architecture chart take default values, points on the processing path corresponding to subparameters in the control parameters are set to normalized values of the subparameters, and 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 plurality of elasticity-imparting process architecture charts into the indicator prediction model to obtain the indicator prediction result, Extracting features within a first feature extraction range in the multiple elasticity-granted process architecture charts based on a short-term feature extraction module constructed by an attention mechanism in the indicator prediction model to obtain short-term features, wherein the first feature extraction range includes elasticity-granted process architecture charts at n time points centered on the elasticity-granted process architecture chart at the target time point, where n is a positive integer, and the short-term feature extraction module uses the elasticity-granted process architecture chart at the target time point as a query feature and the elasticity-granted 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; Extracting features within a second feature extraction range in the elasticity-granting process architecture charts based on a long-term feature extraction module constructed by an attention mechanism in the indicator prediction model to obtain long-term features, where the second feature extraction range includes m time points of elasticity-granting process architecture charts 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 uses a sliding window mechanism to divide the elasticity-granting process architecture charts at the m time points, determine pooling information within each sliding window, and construct key features and value features required for an attention mechanism based on the pooling information of the multiple sliding windows, to obtain the long-term features based on the key features, the value features, and the query features; performing a fusion operation on the long-term features and the short-term features using a fusion module to obtain fused features; Processing the fusion features according to a prediction module in an index prediction model to obtain the index prediction result; The method of claim 2 , comprising:

5. The indicator prediction model further includes a supplemental feature extraction module; The method further comprises: According to the supplementary feature extraction module, perform the following operations to obtain supplementary features: The operation is performing a data analysis on the second control parameter sequence to obtain a cumulative difference of parameters for each time point in the second control parameter sequence relative to a first time point in the second control parameter sequence to obtain a cumulative difference sequence; determining a time length relative to the first time point at which the cumulative difference is greater than a predetermined threshold based on the cumulative difference sequence to obtain a time length sequence; performing feature extraction on the second control parameter sequence, the cumulative difference sequence, and the time length sequence to obtain the supplemental features; Including, performing a fusion operation on the long-term features and the short-term features using the fusion module to obtain a fused feature; The method of claim 4 , further comprising: performing a fusion operation on the long-term features, the short-term features, and the supplemental features using a fusion module to obtain the fused features.

6. The method according to any one of claims 1 to 5, wherein the index prediction results include at least one of grade, pass rate, full winding rate, and dyeing uniformity of the wound yarn package.

7. 1. A method for training a control network for an elasticization process, comprising: the control network includes a parameter prediction sub-network, an indicator prediction sub-network, and a data generation sub-network; The method comprises: obtaining a first training sample comprising predetermined control parameters at a plurality of time points of the elasticization process; inputting the first training sample into the parameter prediction sub-network to obtain parameter predictions for at least one time point in the future; processing the parameter predictions based on the indicator prediction sub-network to obtain predicted indicators; determining a training loss based on the difference between the predicted indicators and the actual indicators and the difference between the parameter predictions and the actual parameter values; adjusting the parameter prediction sub-network and the indicator prediction sub-network based on the training loss; If a training convergence condition is satisfied, obtaining a parameter prediction model corresponding to the parameter prediction sub-network and an indicator prediction model corresponding to the indicator prediction sub-network; training the data generation sub-network based on the parameter prediction model and the indicator prediction model to obtain a data generation model; Including, The method for training a control network for an elasticization process, wherein the data generation model is for generating desired control parameters for an elasticization machine based on the index prediction results output from the index prediction model.

8. the data generation sub-network includes a training wait generator and a training wait classifier; training the data generation sub-network based on the parameter prediction model and the indicator prediction model to obtain a data generation model, training the data-generating network multiple times; Each training session involves the following operations: The operation is inputting a true control parameter into the training-waiting classifier to obtain a first classification result for the true control parameter by the training-waiting classifier; generating fake control parameters using the trained wait generator; inputting the false control parameters into the training classifier to obtain a second classification result; adjusting a model parameter of the training-waiting classifier based on the first discrimination result and the second discrimination result; Fixing model parameters of the training-wait classifier and training the training-wait generator; The method of claim 7, comprising:

9. A control device for an elasticity imparting process, comprising: a collection unit for sequentially collecting predetermined control parameters at multiple time points in the elasticity imparting process flow to obtain a first sequence of control parameters; a parameter prediction unit for inputting the first control parameter sequence into a parameter prediction model to predict predicted parameters for a plurality of future time points; a selection unit for selecting a target parameter for a target time point from the predicted parameters for the plurality of time points; a construction unit for constructing a second control parameter sequence including the target parameter based on control parameters at a plurality of time points before and after the target time point; an index prediction unit for processing the second control parameter sequence based on an index prediction model to obtain an index prediction result; and a data generation unit for inputting the index prediction result into a data generation model to obtain desired control parameters for the target time point if the index prediction result does not satisfy a desired value.

10. The indicator prediction unit: a generating subunit for generating a corresponding elasticity-imparted process architecture chart according to the control parameters of each time point of the second control parameter sequence, thereby obtaining a plurality of elasticity-imparted process architecture charts; a prediction subunit for inputting the plurality of elasticity-imparting process architecture charts into the indicator prediction model to obtain the indicator prediction result; 10. The apparatus of claim 9, comprising:

11. The generating subunit is used to generate a corresponding elasticity-imparting process architecture chart by respectively performing the following operations for the control parameters at each time point of the second control parameter sequence: The operation is performing a normalization operation on each of a plurality of sub-parameters of the control parameter, mapping the sub-parameters to a predetermined value interval, and obtaining normalized values of the sub-parameters; and scaling the processing path in the elasticizing process of the POY yarn to the initialization image in an equal proportion to obtain an elasticizing process architecture chart corresponding to said time point; The apparatus of claim 10, wherein values of pixel points other than the processing path in the elasticity-imparting process architecture chart take default values, points on the processing path corresponding to sub-parameters in the control parameters are set to normalized values of the sub-parameters, and points on the processing path other than the sub-parameters in the control parameters are set to target values different from the default values.

12. The prediction subunit is used to perform the following operations: The operation is Extracting features within a first feature extraction range in the plurality of elasticity-granted process architecture charts based on a short-term feature extraction module constructed by an attention mechanism in the indicator prediction model to obtain short-term features, wherein the first feature extraction range includes elasticity-granted process architecture charts at n time points centered on the elasticity-granted process architecture chart at the target time point, where n is a positive integer, and the short-term feature extraction module uses the elasticity-granted process architecture chart at the target time point as a query feature and the elasticity-granted 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; Extracting features within a second feature extraction range in the elasticity-granting process architecture charts based on a long-term feature extraction module constructed by an attention mechanism in the indicator prediction model to obtain long-term features, where the second feature extraction range includes m time points of elasticity-granting process architecture charts 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 uses a sliding window mechanism to divide the elasticity-granting process architecture charts at the m time points, determine pooling information within each sliding window, and construct key features and value features required for an attention mechanism based on the pooling information of the multiple sliding windows, to obtain the long-term features based on the key features, the value features, and the query features; performing a fusion operation on the long-term features and the short-term features using a fusion module to obtain fused features; and processing the fused features based on a prediction module in an index prediction model to obtain the index prediction result.

13. The indicator prediction model further includes a supplemental feature extraction module; The prediction subunit is further used to perform the following processing: The process comprises: Performing the following operations based on the supplementary feature extraction module to obtain supplementary features; using a fusion module to perform a fusion operation on the long-term features, the short-term features, and the supplemental features to obtain the fused features; The operation is performing a data analysis on the second control parameter sequence to obtain a cumulative difference of parameters for each time point in the second control parameter sequence relative to a first time point in the second control parameter sequence to obtain a cumulative difference sequence; determining a time length relative to the first time point at which the cumulative difference is greater than a predetermined threshold based on the cumulative difference sequence to obtain a time length sequence; and performing feature extraction on the second control parameter sequence, the cumulative difference sequence, and the time length sequence to obtain the supplemental features.

14. The apparatus according to any one of claims 9 to 13, wherein the index prediction results include at least one of grade, pass rate, full winding rate, and dyeing uniformity of the wound yarn package.

15. 1. A training device for a control network of an elasticization process, comprising: the control network includes a parameter prediction sub-network, an indicator prediction sub-network, and a data generation sub-network; The training device comprises: an acquisition unit for acquiring a first training sample comprising predetermined control parameters at multiple time points of the elasticization process; an input unit for inputting the first training sample into the parameter prediction sub-network to obtain parameter predictions for at least one time point in the future; a processing unit for processing the parameter predictions based on the indicator prediction sub-network to obtain a predicted indicator; a training loss unit for determining a training loss based on the difference between the predicted indicators and the actual indicators and the difference between the parameter predicted values and the actual parameter values; an adjustment unit for adjusting the parameter prediction sub-network and the indicator prediction sub-network based on the training loss; a first generation unit for obtaining a parameter prediction model corresponding to the parameter prediction sub-network and an indicator prediction model corresponding to the indicator prediction sub-network when a training convergence condition is met; a second generating unit for training the data generating sub-network based on the parameter prediction model and the indicator prediction model to obtain a data generating model; The data generation model generates desired control parameters for an elasticity applying machine based on the index prediction results output from the index prediction model, thereby providing a training device for a control network of an elasticity applying process.

16. the data generation sub-network includes a training wait generator and a training wait classifier; the second generation unit is used to train the data-generating network multiple times; Each training session involves the following operations: The operation is inputting a true control parameter into the training-waiting classifier to obtain a first classification result for the true control parameter by the training-waiting classifier; generating fake control parameters using the trained wait generator; inputting the false control parameters into the training classifier to obtain a second classification result; adjusting a model parameter of the training-waiting classifier based on the first discrimination result and the second discrimination result; and fixing model parameters of the trained classifier and training the trained generator.

17. An electronic device, at least one processor; a memory communicatively connected to the at least one processor; An electronic device, wherein a memory has stored therein instructions executable by at least one processor, the instructions being executed by the at least one processor to cause the at least one processor to perform the method of any one of claims 1 to 5.

18. A non-transitory computer readable storage medium having stored thereon computer instructions, the computer instructions being used to cause a computer to perform the method of any one of claims 1 to 5.

Images