Elasticity enhancing control method, device and equipment based on false twisting component and storage medium

Abstract

The invention provides an elasticizing control method, device and equipment based on a false twisting component and a storage medium, and relates to the technical field of computers. According to the specific implementation scheme, the method comprises the following steps: inputting a video segment of a current time window in a video stream collected by a false twisting part into a visual encoder to obtain an initial visual feature; inputting a parameter sequence of active control parameters and passive monitoring parameters, belonging to the current time window, of the elasticizer where the false twisting component is located into a text encoder to obtain text features; inputting the initial visual features and the text features into a large language model to construct key value pair information of at least one decoding layer of the large language model; optimizing a visual feature representation in the key value pair information based on the controller; on the basis of the optimized visual feature representation, a detection result of the false twisting part is predicted, and the detection result at least comprises the abrasion degree and the remaining life; and controlling the control parameters of the elasticizing process based on the detection result.

Description

Technical Field

[0001] This disclosure relates to the field of computer technology, and in particular to the technical fields of large models, neural networks, and texturing control methods for false twist components. Background Technology

[0002] In the chemical fiber textile industry, the texturing machine is the core equipment for converting raw yarn into elastic yarn, and the false twisting component is a core component of the texturing machine. The false twisting component applies periodic false twisting to the yarn through high-speed rotation, which directly determines key quality indicators such as twist uniformity, tension stability, and elastic recovery rate of the finished yarn.

[0003] Therefore, the matching degree between the operating status of the false twist component and the texturing process not only affects the yarn production efficiency, but is also directly related to production losses such as yarn breakage rate and downgrade rate. Summary of the Invention

[0004] This disclosure provides a texturing control method, apparatus, device, and storage medium based on a false twist component to solve or alleviate one or more technical problems in the prior art.

[0005] According to one aspect of this disclosure, a texturing control method based on a false twist component is provided, comprising: The video segment of the current time window in the video stream acquired from the false twist component is input into the visual encoder to obtain the initial visual features; Input the parameter sequence of active control parameters and passive monitoring parameters of the texturing machine where the false twist component is located, which belongs to the current time window, into the text encoder to obtain text features; Initial visual and textual features are input into a large language model to construct key-value pair information for at least one decoding layer of the large language model; Optimize the visual feature representation in key-value pair information based on the controller; Based on the optimized visual feature representation, the detection results of the false-twist component are predicted, and the detection results include at least the degree of wear and the remaining life. Based on the test results, control parameters for the texturing process are adjusted.

[0006] According to another aspect of this disclosure, a texturing control device based on a false twist component is provided, comprising: The first input module is used to input the video segment of the current time window in the video stream acquired from the false twist component into the visual encoder to obtain the initial visual features; The second input module is used to input the parameter sequence of active control parameters and passive monitoring parameters of the texturing machine where the false twist component is located, which belongs to the current time window, into the text encoder to obtain text features; The third input module is used to input the initial visual features and text features into the large language model to construct key-value pair information for at least one decoding layer of the large language model; The optimization module is used to optimize the visual feature representation in key-value pair information based on the controller. The prediction module is used to predict the detection results of the false-twist component based on the optimized visual feature representation. The detection results include at least the degree of wear and the remaining life. The control module is used to control the parameters of the texturing process based on the detection results.

[0007] According to another aspect of this disclosure, an electronic device is provided, comprising: At least one processor; and The memory is communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform the methods of any embodiment of the present disclosure.

[0008] According to another aspect of this disclosure, a non-transitory computer-readable storage medium is provided storing computer instructions, wherein the computer instructions are used to cause the computer to perform a method according to any embodiment of this disclosure.

[0009] According to another aspect of this disclosure, a computer program product is provided, including a computer program that, when executed by a processor, implements a method according to any embodiment of this disclosure.

[0010] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of this disclosure, nor is it intended to limit the scope of this disclosure. Other features of this disclosure will become readily apparent from the following description. Attached Figure Description

[0011] In the accompanying drawings, unless otherwise specified, the same reference numerals throughout the various drawings denote the same or similar parts or elements. These drawings are not necessarily drawn to scale. It should be understood that these drawings depict only some embodiments provided according to this disclosure and should not be construed as limiting the scope of this disclosure.

[0012] Figure 1 This is a schematic diagram of a texturing machine according to the first embodiment of this disclosure; Figure 2 This is a flowchart illustrating the texturing control method based on a false twist component according to the second embodiment of this disclosure; Figure 3 This is a schematic flowchart illustrating the visual feature representation in key-value pair information based on the controller optimization according to the third embodiment of this disclosure; Figure 4 This is a schematic diagram illustrating the determination of the initial values ​​of the controller parameters according to the fourth embodiment of this disclosure; Figure 5 This is a schematic diagram of the texturing control device based on a false twist component according to the eighth embodiment of this disclosure; Figure 6 This is a schematic diagram of an electronic device structure for a texturing control method based on a false twist component according to the ninth embodiment of this disclosure. Detailed Implementation

[0013] The present disclosure will now be described in further detail with reference to the accompanying drawings. The same reference numerals in the drawings denote elements that have the same or similar functions. Although various aspects of embodiments are shown in the drawings, they are not necessarily drawn to scale unless specifically indicated otherwise.

[0014] Furthermore, to better illustrate this disclosure, numerous specific details are set forth in the following detailed description. Those skilled in the art will understand that this disclosure can be practiced without certain specific details. In some instances, methods, means, components, and circuits well known to those skilled in the art have not been described in detail in order to highlight the main points of this disclosure.

[0015] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this disclosure, "multiple" means two or more, unless otherwise explicitly specified.

[0016] It should be noted that, unless it is explicitly stated that there is a sequential order of execution between different operations, or that there is a sequential order of execution between different operations in terms of technical implementation, the execution order between multiple operations may not be significant, and multiple operations may be executed simultaneously.

[0017] Texturing machines are core equipment used in the chemical fiber textile industry to stretch, false-twist, and heat-set raw yarns to produce elastic yarns with elasticity and fluffiness.

[0018] A schematic diagram of a possible texturing machine is shown below. Figure 1 As shown, the key components of the texturing machine include: a raw yarn frame 101, a yarn cutter 102, a first roller 103, a texturing heat box 104, a cooling plate 105, a false twisting component 106, a nozzle 107, a second roller 108, a shaping heat box 109, a third roller 110, a yarn breakage detection device 111, and a winding component 112.

[0019] The first roller 103, the second roller 108, and the third roller 110 are used to ensure that the silk threads are processed along a predetermined path. In the texturing process, the speeds of the first roller 103, the second roller 108, and the third roller 110 are matched to ensure that the silk fabric is not broken or piled up.

[0020] Depending on the product form requirements, the nozzle 107 can be used to process the yarn that has been treated by the false twisting component into a network form of yarn to ensure the required feel and form of different yarns.

[0021] When producing high-elasticity yarn, the setting box 109 is not used. When producing medium-elasticity yarn, the temperature of the setting box 109 can be adjusted to a first preset temperature, for example, around 140°C. When producing low-elasticity yarn, the temperature of the setting box 109 can be adjusted to a second preset temperature, for example, between 165°C and 195°C.

[0022] If the breakage detection device 111 detects a breakage, it will trigger the filament cutter 102 to cut the filament, thus preventing the POY filament from accumulating in subsequent processes.

[0023] In traditional texturing processes, the monitoring and control of false-twist components mainly rely on manual experience and simple physical sensors. Manual inspection is not only inefficient but also fails to capture subtle changes in the false-twist components in real time and accurately, easily leading to delayed fault detection and impacting product quality and production schedules. Furthermore, simple physical sensors can only acquire limited physical parameters, such as temperature and vibration, and cannot comprehensively and intuitively reflect the actual working state of the false-twist components, such as surface wear and internal microscopic damage.

[0024] As false twist components are used continuously, their wear gradually increases. When the wear reaches a certain level, it can lead to instability in the texturing process, resulting in uneven yarn tension and twist, which seriously affects the quality of the texturized yarn. Furthermore, the remaining lifespan of false twist components is difficult to predict accurately. To avoid production interruptions due to sudden component failure, companies often adopt a strategy of periodically replacing false twist components. However, this approach may waste resources and increase production costs.

[0025] In view of this, this disclosure proposes a texturing control method based on a false twist component, which predicts and adjusts the control parameters of the texturing process to solve at least one of the above-mentioned problems.

[0026] It should be noted that the main types of spinning products involved in the embodiments of this disclosure may include one or more of pre-oriented yarns (POY), fully drawn yarns (FDY), and polyester staple fiber. For example, the specific type of yarn may include polyester pre-oriented yarns (POY), polyester fully drawn yarns (FDY), polyester staple fiber, etc.

[0027] In addition, spinning products can also include polyester chips (PET), polycaprolactam chips (PA6), and other chip products.

[0028] like Figure 2 The diagram shown is a flowchart illustrating the texturing control method based on a false-twist component provided in this disclosure, including the following: S201, input the video segment of the current time window in the video stream acquired from the false twist component into the visual encoder to obtain the initial visual features.

[0029] A video stream captured from a false twist component refers to a sequence of dynamic images generated by continuously capturing the real-time operation of the false twist component using visual acquisition devices such as industrial cameras and high-speed video cameras. This video stream can completely record information such as the operating posture, surface condition, and yarn movement trajectory of the false twist component.

[0030] A time window is a fixed-duration video segment extracted from a continuous video stream. The duration of the time window can be set according to the detection requirements.

[0031] A visual encoder is a feature extraction module built on deep learning or traditional machine vision algorithms. Its core function is to extract and transform features from image information in a video through a neural network model, converting video data into a high-dimensional vector with representational characteristics that a computer can understand and process.

[0032] During implementation, a video segment within the current time window can be selected from the video stream of the false twist component and input into the visual encoder. The visual encoder then extracts features from the input video segment to obtain initial visual features.

[0033] S202, input the parameter sequence of active control parameters and passive monitoring parameters of the texturing machine where the false twist component is located, which belongs to the current time window, into the text encoder to obtain text features.

[0034] Active control parameters refer to process parameters that can be actively adjusted and set by operators based on factors such as production needs, process standards, and equipment operating status. These parameters directly affect the texturing process and have a decisive impact on the quality of the final finished yarn.

[0035] Passive monitoring parameters refer to parameters that are collected in real time by sensors from the texturing machine and reflect the actual operating status of the texturing machine and the actual situation of the yarn processing process.

[0036] During implementation, the ordered data set obtained by continuously sampling the active control parameters and passive monitoring parameters within the current time window can be used as the parameter sequence for the current time window.

[0037] It is understandable that during the video stream acquisition process, the acquisition of timestamps, active control parameters, and passive monitoring parameters also has corresponding timestamps. By aligning the timestamps, video segments, active control parameters, and passive monitoring parameters belonging to the same time window can be identified.

[0038] A text encoder is a feature extraction module that can map a structured sequence of parameters into a high-dimensional semantic vector. It extracts features from text data and converts the sequence of parameters into a text feature vector.

[0039] In practice, the obtained parameter sequence can be input into the text encoder, and the high-dimensional vector output by the text encoder is the corresponding text feature.

[0040] S203, input the initial visual features and text features into the large language model to construct key-value pair information for at least one decoding layer of the large language model.

[0041] Large Language Models (LLMs) are deep learning models trained on large amounts of text data that can generate natural language text or understand the meaning of language text. LLMs can handle a variety of natural language tasks, such as text classification, question answering, and dialogue.

[0042] In practice, initial visual features and text features can be input into a large language model. The attention module of each decoding layer will calculate and generate corresponding Key and Value based on these input features to build and pre-populate key-value pair cache information of at least one decoding layer.

[0043] Furthermore, in the last decoding layer, the value cache part corresponding to the visual features is the highest-order semantic representation of the visual features after processing by the entire network, which directly affects the final prediction result.

[0044] S204, Optimize the visual feature representation in key-value pair information based on the controller.

[0045] In this embodiment of the disclosure, the controller is a trainable low-rank parameter matrix, which is the core control unit used to optimize the value cache corresponding to visual features in the decoding layer of a large language model. Its dimension matches the value cache corresponding to the visual features (the highest-order semantic representation of the visual features after processing by the full-layer network), so that the visual feature representation can be iteratively optimized through the parameters of the controller to improve the accuracy of the final prediction result.

[0046] In practice, the controller can be set to a zero matrix in its initial state, and then the parameters of the controller can be updated through a loss function to optimize the visual feature representation in the key-value pair information.

[0047] S205, based on the optimized visual feature representation, predicts the detection results of false-twist components, including at least the degree of wear and remaining life.

[0048] After obtaining the optimized visual feature representation, the subsequent neural network layers of the large language model will perform a prediction process based on the visual feature representation to establish a mapping relationship between the visual features and the wear degree and remaining life of the false twist component, and then output the corresponding detection results.

[0049] S206, based on the test results, control parameters for the texturing process.

[0050] Based on the predicted wear and remaining life of the false twist component, the control parameters of the texturing process can be adjusted. For example, if the test results show that the false twist component is severely worn and has a short remaining life, it can be promptly replaced to ensure the smooth operation of subsequent production.

[0051] In this embodiment, by fusing the visual and textual features of the false-twist component, a multimodal key-value pair feature is constructed using a large language model. The visual feature representation is then optimized by a controller, enabling accurate and real-time prediction of the wear level and remaining life of the false-twist component. Based on the prediction results, the texturing process control parameters are dynamically adjusted. This solves the problem of low efficiency in manual inspection. Leveraging the learning and reasoning capabilities of a large model, it can detect subtle faults that are difficult to detect in a timely manner, thereby improving production quality and stability.

[0052] This invention leverages the powerful learning capabilities and prior knowledge of the model, combined with parameters collected from vision and multiple sensors, to comprehensively analyze and detect the state of the twisting component. This overcomes the limitations of a single sensor in fully reflecting the actual working conditions of the component, such as surface wear and microscopic damage. It effectively reduces product quality issues caused by component abnormalities, such as uneven yarn tension and twist, thereby ensuring smooth and stable production. Furthermore, the detection results of false-twist components can replace the traditional periodic replacement strategy, reducing unnecessary component replacement frequency, lowering resource waste and production costs, and ultimately achieving intelligent and precise control of the texturing process.

[0053] In this embodiment of the disclosure, the active control parameters include at least one of the following: (1) False twist ratio; This refers to the ratio of the disc rotation speed within the false twist component to the machine speed. The machine speed refers to the speed of the second roller.

[0054] (2) Box temperature; The heating chamber is a component in a texturing machine used to heat and shape the yarn. The heating chamber temperature is the internal heating temperature of the heating chamber. Specifically, the heating chamber temperature can include the temperature of the forming heating chamber and the temperature of the setting heating chamber.

[0055] (3) Turbine speed.

[0056] The oil roller is a component in a texturing machine used to apply oil to the yarn. The oil roller speed refers to the rotational speed of this component.

[0057] The passive monitoring parameters include at least one of the following: (1) Thread tension; Thread tension refers to the magnitude of the tensile force exerted on the thread during the texturing process.

[0058] (2) Decapitation rate; The breakage rate refers to the ratio of the number of times a wire breaks in a texturing machine to the total number of wires processed within a certain period of time.

[0059] (3) Continuous operating time of the false-twist component; The continuous running time of the false twist component refers to the cumulative running time of the false twist component from the start of operation to the current moment.

[0060] (4) Bearing temperature of the false-twisted component; The bearing temperature of a false-twist component refers to the operating temperature of the bearing within the false-twist component.

[0061] (5) Historical operation records of the texturing machine.

[0062] The historical operation record of the texturing machine refers to the record of information such as the operating parameters, fault conditions, and production output of the texturing machine over a period of time.

[0063] In this embodiment, by integrating active control parameters and passive monitoring parameters, the operating status of the texturing machine containing the false-twist component can be reflected from multiple dimensions and perspectives. This provides rich and accurate information for subsequent large-scale language model prediction of the false-twist component's detection results. Ultimately, this helps to more accurately grasp the wear level and remaining life of the false-twist component, thereby enabling more reasonable control of the texturing process's control parameters.

[0064] In this embodiment of the disclosure, the visual feature representation in key-value pair information is optimized based on the controller. For example... Figure 3 As shown, for each target decoding layer in at least one decoding layer, the following operations can be performed iteratively: S301, For the current step, obtain the input information of the target decoding layer. The input information includes the current output result of the large language model and the current key-value pair information of the target decoding layer.

[0065] Each round of optimization is considered as one step, and the current step refers to the current round of optimization.

[0066] Among them, the current output of the large language model is the current detection result output by the large language model for the false twist component.

[0067] S302 uses a large language model to predict the input information and obtains the confidence level of the intermediate output result of the current step.

[0068] The intermediate output results are also the non-final detection results of the large language model output layer for the current time window. This disclosure optimizes the detection results of the large language model for false twist components by continuously optimizing the visual feature representation through an optimizer.

[0069] In implementation, after receiving input information, the large language model can first use its own attention mechanism and the operations of the decoding and output layers to perform feature extraction and association analysis on the input information, and calculate the logits corresponding to the next label in the current step. Among them, logits are the unnormalized raw prediction scores, which directly reflect the degree of preference of the large language model for candidate results of each category.

[0070] Next, check logits Performing Softmax normalization transforms the raw scores into a probability distribution that conforms to probability axioms. Each element in the probability distribution , which is the confidence level of the intermediate output result of the current step, and its value represents the reliability of the large language model in determining that the current monitoring data belongs to the i-th type (fault or normal state).

[0071] S303, determine the distribution entropy of the current step based on the confidence level.

[0072] Distribution entropy is an index used to quantify the uncertainty of an intermediate output result, calculated based on the probability distribution of the confidence level of the intermediate output result at the current step. The larger the distribution entropy, the more dispersed the probability distribution of the confidence level, and the higher the uncertainty of the intermediate result; the smaller the distribution entropy, the more concentrated the probability distribution of the confidence level, and the stronger the certainty of the intermediate output result.

[0073] During implementation, the distribution entropy of the current step is determined based on the confidence level, which can be described by expression (1): (1) In expression (1), V represents the distribution entropy at time t, which is also the distribution entropy at the current step; V represents the set of all possible values ​​of the confidence score of the intermediate output result at the current step. This indicates the confidence level of the intermediate output result in the current step; The logarithm represents the confidence level of the intermediate output of the current step, used to achieve a reasonable quantification of uncertainty.

[0074] S304. Determine the exponential moving average entropy of the current step based on the distribution entropy and the exponential moving average entropy of the previous step.

[0075] Exponential Moving Average Entropy (EMA) is a statistical indicator that combines the exponential moving average (EMA) with information entropy. By giving more weight to recent data and reducing the impact of historical data fluctuations, this indicator more accurately reflects the changing trend of entropy values.

[0076] During implementation, the exponential moving average entropy of the current step is determined and can be described by expression (2): (2) In expression (2), This represents the exponential moving average entropy of the current step; This represents the exponential moving average entropy of the previous step in the current step. This represents the distribution entropy of the current step; This represents the smoothing coefficient, with a value range of [0,1]. The closer the value is to 1, the higher the weight of the exponential moving average entropy of the previous step in the current step, and the stronger the smoothing effect.

[0077] S305, determine the target loss based on the exponential moving average entropy of the current step.

[0078] The target loss is the core indicator for measuring the degree of deviation between the model's prediction results and the ideal state; the magnitude of the target loss directly guides the optimization direction of the controller parameters.

[0079] During implementation, the target loss can be described by expression (3): (3) In expression (3), Indicates target loss; This represents the weighting coefficient, which takes a value of 1 or -1 and is used to guide the optimization direction. This represents the exponential moving average entropy of the current step. The controller parameters are optimized by minimizing the target loss, thereby improving the visual feature representation.

[0080] S306, Determine the gradient of the target loss with respect to the controller parameters.

[0081] The gradient is the partial derivative of the target loss function with respect to the controller parameters, reflecting the rate and direction of change of the target loss with respect to the controller parameters; the sign and magnitude of the gradient determine the direction of the controller parameter update.

[0082] In practice, the gradient of the target loss with respect to the controller parameters is determined, and can be described by expression (4): (4) In expression (4), The loss function represents the loss function with respect to the controller parameters. The gradient; Represents the target loss function of the controller; This represents the distribution entropy of the current step; Represents the weight coefficients of the k-th iteration; in At this time, the gradient direction is the opposite of the decreasing direction of the distributed entropy, driving the controller parameters to adjust in the direction of increasing distributed entropy; When the gradient direction is in the same direction as the decreasing direction of the distributed entropy, the controller parameters are adjusted in the direction of decreasing distributed entropy.

[0083] S307 updates the controller parameters based on gradients to obtain the controller parameter update results.

[0084] That is, based on the gradient calculated in step S306, the controller parameters are updated by using an optimization algorithm (such as stochastic gradient descent) to obtain the updated controller parameters.

[0085] In practice, the controller parameters can be adjusted using an optimizer (such as AdamW), and the number of parameters can be compressed using a low-rank matrix to reduce computational complexity and speed up feature extraction.

[0086] During implementation, update the controller parameters to obtain an update result, which can be described by expression (5): (5) In expression (5), represents the updated controller parameters; represents the current controller parameters; represents the learning rate, which controls the optimization step size; represents the gradient of the loss function with respect to the controller parameters .

[0087] Among them, the number of controller parameters is compressed in the form of a low-rank matrix, which can be described by expression (6): (6) In expression (6), represents the original adjustable parameters of the controller, and are the trainable weight matrices inside the controller, ; among them, r << d ensures a significant compression of the number of parameters to reduce the computational amount while avoiding model overfitting; represents the visual features; represents the scaling factor.

[0088] S308, update the visual feature representation in the current key-value pair information based on the controller parameter update result.

[0089] That is, use the updated controller parameters to update the visual feature representation in the current key-value pair information.

[0090] During implementation, updating the visual feature representation in the current key-value pair information based on the controller parameter update result can be expressed by expression (7): (7) In expression (7), represents the updated visual feature representation in the current key-value pair information; represents the original visual feature representation, that is, the visual features stored in the key-value pair before update; represents the visual feature representation adjusted based on the control parameters; represents the norm of the adjusted visual feature representation; represents the norm of the original visual feature representation, which is used to keep the feature energy of the updated visual features consistent with the original visual features and avoid affecting the stability of subsequent key-value pair matching or fusion calculations due to feature scale fluctuations; among them, the feature energy is a quantitative description of the information-bearing intensity and numerical scale of the visual feature vector.

[0091] S309, if it is determined that the termination condition is not met, construct the input information for the next step of the current step based on the updated visual feature representation, take the next step as the current step, and return to execute the operation of obtaining the input information of the target decoding layer for the current step.

[0092] That is, after completing an update, check whether the termination condition is met. The termination condition may be that a preset number of iterations has been reached, or the target loss is less than a certain threshold. If the termination condition is not met, construct the input information for the next step based on the updated visual feature representation, take the next step as the current step, and then return to execute the operation of obtaining input information to continue iterative optimization.

[0093] In this embodiment of the disclosure, the distribution entropy is calculated by confidence level and the target loss is constructed by combining it with the exponential moving average entropy. Then, the controller parameters are optimized based on the loss gradient, and the visual feature representation in the key-value pairs is dynamically adjusted. This can continuously reduce the uncertainty of the model in predicting the state of the false-twisted component, and ultimately improve the accuracy and stability of the wear degree and remaining life prediction.

[0094] In this embodiment of the disclosure, the initial values ​​of the controller parameters are determined, such as... Figure 4 As shown, it includes the following: S401, acquire the reference video sequence and reference parameter sequence of the false twist component under normal working conditions.

[0095] A reference video sequence is a series of video frames recording the false-twist component in its normal operating state. These video frames may contain visual information such as the appearance, movement, and material morphology of the false-twist component. For example, texture information of the false-twist component in a non-wear-prone state.

[0096] The reference parameter sequence records the active control parameters and passive monitoring parameters of the false twist component during normal operation.

[0097] S402, perform principal component analysis on the reference video sequence to obtain the initial parameters of the first preset channel set of the controller.

[0098] Principal component analysis (PCA) is an unsupervised technique for dimensionality reduction and feature extraction. It transforms multiple correlated variables in the original high-dimensional data into a set of uncorrelated new variables, known as principal components. These principal components are then sorted according to their variances from largest to smallest. Principal components with larger variances contain more information from the original data. Selecting the first few principal components with the largest variances can achieve dimensionality reduction while preserving the core features of the original data.

[0099] During implementation, principal component analysis is performed on the reference video sequence to extract the principal component basis vectors representing the visual patterns in the normal state. These basis vectors are then filled into the first preset channel set of the controller to complete the parameter initialization of the visual feature channels. For example, high-dimensional visual features are compressed into a 16-dimensional latent space to focus on feature patterns strongly correlated with the wear of false-twist components.

[0100] S403, perform principal component analysis on the reference parameter sequence to obtain the initial parameters of the second preset channel set of the controller.

[0101] Principal component analysis is performed on the reference parameter sequence to extract principal component basis vectors representing the parameter patterns of the normal state. These basis vectors are then filled into the second preset channel set of the controller to complete the parameter initialization of the parameter feature channels. For example, the initial parameters are compressed into a 4-dimensional latent space, including both active control parameters and passive monitoring parameters, to ensure that there are no anomalies in the initial state.

[0102] In this embodiment of the disclosure, by performing principal component analysis on the reference video sequence and reference parameter sequence of the false twist component under normal working conditions, the initial parameters of the first preset channel set and the initial parameters of the second preset channel set of the controller are obtained. This enables the controller to converge to the appropriate parameter settings more quickly and accurately based on the characteristics of the false twist component under normal working conditions when it begins to optimize the visual feature representation in the key-value pair information. This improves the accuracy of subsequent prediction of the false twist component detection results, thereby more effectively controlling the control parameters of the texturing process and ensuring the stability and efficiency of the texturing process.

[0103] In this embodiment, when the remaining lifespan of the false-twist component is below the lifespan threshold, the false-twist component may be approaching a critical state of severe wear or even failure, resulting in a significant decrease in overall operational stability. To avoid fluctuations in the texturing process caused by frequent iterations of controller parameters when the false-twist component is nearing the end of its lifespan, [further measures are needed].

[0104] During implementation, if the remaining life of the false twist component predicted in multiple consecutive time windows is less than the life threshold, the fixed controller parameters will no longer change.

[0105] Among them, the life threshold is a pre-set standard value that represents the critical value of the remaining life of the false twist component.

[0106] During implementation, machine learning algorithms can be used to predict the remaining lifespan of the false-twist component within the current time window, and prediction data from multiple consecutive time windows can be continuously collected. If it is determined that the remaining lifespan of the false-twist component predicted in multiple consecutive time windows is less than the lifespan threshold, a controller parameter freezing mechanism is triggered: the core parameters of the controller are fixed, and the gradient iteration update process of the parameters is terminated.

[0107] In this embodiment, by fixing the controller parameters, fluctuations in the texturing process caused by frequent iterations of the controller parameters can be avoided, maintaining a relatively stable predictive state. Simultaneously, it provides a stable basis for timely judgment to take maintenance measures such as replacing components, ensuring the continuity and stability of the texturing process and avoiding unnecessary interference and losses to the entire texturing production process due to frequent parameter changes.

[0108] In this embodiment, control parameters of the texturing process are controlled based on the detection results. Based on the wear degree and remaining life of the false-twist component, when the false-twist component is in a first state, the false-twist ratio and the temperature of the first hot box are adjusted based on the wear degree; wherein, the false-twist ratio is negatively correlated with the wear degree; the hot box temperature is increased in specified steps, and the adjusted hot box temperature is lower than the upper temperature limit; wherein, the first state implicitly indicates that the wear amount of the false-twist component is lower than the target amount; the first hot box is... Figure 1 The deformable heating box 104 shown in the figure has a heating temperature that can soften the yarn.

[0109] Implicit representation is a method of representing the state indirectly by means of associated detectable features, rather than through direct numerical values, identifiers, or visual information. The target quantity is the wear threshold of the false-twist component. In the embodiments of this disclosure, the first state is used to indicate that the false-twist component has no pattern or is in the early stage of wear (i.e., low wear).

[0110] In practice, for example, after replacing a new false twist component, if the wear of the false twist component is below the lower threshold and the remaining lifespan is greater than the upper threshold, the new false twist component is determined to be in the first state within a first time period set according to empirical values.

[0111] The first state can be understood as the early stage of failure, during which the performance of the false-twist component is relatively good. Therefore, the control objective at this stage is to slow down the wear rate of the component and maintain production continuity.

[0112] In practice, the false twist ratio is adjusted based on the degree of wear, which can be described by expression (8): (8) In expression (8), Indicates the false twist ratio; This indicates the degree of wear. As the wear of the false-twist components increases, the false-twist ratio decreases accordingly; conversely, when the wear is minimal, the false-twist ratio can be appropriately increased. This is a preset constant, with a value less than 1.

[0113] During the adjustment process, the false twist ratio must meet the physical boundary constraints to prevent insufficient yarn twist due to an excessively low false twist ratio, which would affect product quality.

[0114] In practice, when the false twist component is in its first state, the oven temperature can be gradually increased in a pre-set step size according to its wear level. For example, if the specified step size is m℃, then the oven temperature is increased by m℃ each time it is adjusted. During the adjustment process, to avoid damage to the yarn or affecting the normal operation of the false twist component due to excessive temperature, it must be ensured that the adjusted temperature is lower than the pre-set upper temperature limit. The upper temperature limit can be determined based on factors such as the yarn material and the heat resistance of the false twist component.

[0115] In addition, during the process of adjusting the false twist ratio and the temperature of the first heating box, the yarn breakage rate corresponding to each set of adjusted process parameters must be observed and recorded simultaneously.

[0116] In this embodiment, when the wear level and remaining life of the false twist component indicate that it is in a first state, the false twist ratio and the temperature of the first heating box are adjusted according to the wear level. The false twist ratio is negatively correlated with the wear level, thus allowing for reasonable adjustment based on component wear to ensure the twist effect during yarn processing. Simultaneously, the heating box temperature is increased in specified steps, but ensured to remain below the upper temperature limit. By appropriately increasing the heating box temperature, the yarn is softened better to adapt to the wear condition of the false twist component, maintaining the quality and efficiency of yarn processing, and further ensuring the quality and performance of the yarn after texturing.

[0117] In this embodiment of the disclosure, after determining the gradient of the target loss with respect to the controller parameters, the gradient can be compared with an upper bound on the gradient; if the gradient is greater than the upper bound, the gradient is clipped to the upper bound.

[0118] The gradient upper bound is a pre-defined threshold used to limit the magnitude of the gradient and prevent it from becoming too large. When a gradient is found to be greater than the gradient upper bound, it needs to be clipped. For example, elements in the gradient greater than the gradient upper bound can be directly set to the gradient upper bound, while elements less than the gradient upper bound remain unchanged.

[0119] By clipping the gradient to the upper limit of the gradient, it is possible to effectively avoid drastic fluctuations in the controller parameters during the iterative optimization process, and ensure that the update process of the visual feature representation in the key-value pair converges smoothly. The stable optimization of the visual feature representation will directly improve the accuracy of the prediction results of the wear degree and remaining life of the false twist component, enabling it to accurately determine whether the false twist component is in the first state.

[0120] When the prediction result stably points to the first state, the false twist ratio can be adjusted based on the accurate wear degree with a negative correlation, and the temperature of the first hot box can be safely increased in a specified step to achieve smooth and controllable adjustment of the texturing process parameters. If gradient cutting is lacking, the controller parameters are prone to unstable updates of visual feature representation due to gradient oscillation, which may lead to misjudgment of component status. For example, misjudging a non-first state as the first state and blindly adjusting the process parameters may cause problems such as uneven yarn tension and uneven twist, or missing the first state and missing the opportunity for process optimization, ultimately affecting the quality of texturing products and production stability.

[0121] In this embodiment of the disclosure, by comparing the gradient and the upper limit of the gradient, and if the gradient is greater than the upper limit of the gradient, the gradient is clipped to the upper limit of the gradient, which can effectively control the gradient update process and avoid adverse effects on the update of controller parameters due to excessively large gradients.

[0122] In this embodiment of the disclosure, the control parameters for the texturing process based on the detection results further include: When the false twist component is in the second state, the false twist ratio and the target hot box temperature used in the texturing process when the yarn breakage rate is the lowest within the target time range are obtained and used as the target false twist ratio and the target hot box temperature, respectively. The texturing process is fixed to use the target false twist ratio and the target hot box temperature; The second state is used to implicitly indicate that the wear of the false twist component is greater than or equal to the target amount.

[0123] In practice, the implicit representation here has a similar meaning to the previous one, using related quantities instead of direct quantities to represent the second state.

[0124] For example, when the actual wear of the false-twist component exceeds the preset wear limit, it is determined to be in the second state. The second state can be understood as the late stage of the fault, at which point the wear of the false-twist component is high, its operating performance is significantly reduced, and further parameter adjustments are likely to cause process fluctuations. Therefore, the control objective at this stage is to lock in the optimal compensation parameters and prepare for preventative maintenance.

[0125] The target time range is a predetermined period of time that can be set based on factors such as production plans and equipment operating cycles.

[0126] Based on the above, within the target time range, online data is recorded for the texturing process under different combinations of false twist ratio and target hot box temperature. The set of false twist ratio and hot box temperature that minimizes the yarn breakage rate is identified and defined as the target false twist ratio and target hot box temperature.

[0127] After determining the target false twist ratio and target hot box temperature, the texturing process parameters are fixed at the target false twist ratio and target hot box temperature. The purpose of this is to ensure production stability and product quality as much as possible, and reduce the adverse effects of component wear, when the false twisted parts are in the second state (severe wear), by using process parameters that minimize the breakage rate.

[0128] In this embodiment of the disclosure, when the false twist component is in the second state, it means that the false twist component is severely worn. By adjusting and fixing the texturing process based on the false twist ratio when the yarn breakage rate is the lowest and the hot box temperature of the target hot box, it is possible to avoid the yarn breakage rate from increasing due to arbitrary changes in parameters, thus ensuring the continuity and stability of the production process, improving production efficiency, and also helping to ensure the quality of yarn products and reduce the adverse effects caused by severe wear of the false twist component.

[0129] In this embodiment of the disclosure, the false-twist component is determined to enter the second state when at least one of the following conditions is met: Condition 1: The decrease in the exponential moving average entropy over n consecutive time windows is greater than the target magnitude, and the distribution entropy of the current step is lower than the target percentage of the preset peak value. n is a pre-defined positive integer representing the number of consecutive time windows.

[0130] Based on the preceding explanation, the exponential moving average entropy of the current step is determined by using the distribution entropy and the exponential moving average entropy of the previous step. Within a single time window, multiple exponential moving average entropy values ​​will be obtained due to multiple iterations of the smoothing coefficient calculation. For example, within a single time window, the set {0.82, 0.79, 0.85, 0.77} can be obtained through iteration.

[0131] During implementation, the maximum or minimum value can be selected as the representative entropy value for the window based on the judgment requirements. Alternatively, the arithmetic mean of these values ​​can be calculated as the representative entropy value for the time window.

[0132] During implementation, if the exponential moving average entropy shows a downward trend within n consecutive time windows, and the magnitude of the decrease exceeds the preset target magnitude, it indicates that the operating state of the false twist component has changed significantly over a period of time, and abnormal conditions such as wear may occur.

[0133] The preset peak value is the maximum value of the distribution entropy recorded during the normal operation of the false twist component. If the distribution entropy in the current step is lower than the target percentage of the preset peak value, it indicates that the current distribution entropy has significantly decreased compared to the maximum value during normal operation, further indicating a significant change in the state of the false twist component. The target percentage is a preset ratio, which can be set according to actual needs during implementation; this disclosure does not limit its implementation.

[0134] Therefore, if the decrease in the exponential moving average entropy over n consecutive time windows is greater than the target magnitude, and the distribution entropy of the current step is lower than the target percentage of the preset peak value, it can be determined that the false twist component has entered the second state.

[0135] Condition 2: The exponential moving average entropy of the current step is less than the target threshold determined based on empirical values, and the decrease in the exponential moving average entropy of the current step compared to the preset peak value is greater than the preset magnitude.

[0136] The target threshold is a fixed value determined based on historical experience data. When the exponential moving average entropy of the current step is less than this target threshold, it indicates that the current operating state of the false twist component has deviated from the normal range, and there may be issues such as wear on the false twist component.

[0137] The preset amplitude is a pre-defined decrease value. If the decrease of the previous exponential moving average entropy from the preset peak exceeds this preset amplitude, it indicates that the state of the false twist component has changed drastically, with a significant difference between the normal state and the current state.

[0138] Therefore, if the exponential moving average entropy of the current step is less than the target threshold determined based on empirical values, and the decrease of the exponential moving average entropy of the current step compared to the preset peak value is greater than the preset magnitude, it can be determined that the false twist component has entered the second state.

[0139] In this embodiment of the disclosure, condition 1 involves the exponential moving average entropy decreasing by a greater than the target magnitude over n consecutive time windows, and the current step distribution entropy being lower than the target percentage of the preset peak value. Condition 2 involves the current step exponential moving average entropy being less than the target threshold determined based on empirical values, and the decrease compared to the preset peak value being greater than the preset magnitude. By setting and judging these two conditions, the operating status of the false twist component can be comprehensively evaluated from different perspectives, improving the accuracy and reliability of the judgment. This allows for timely implementation of corresponding process adjustment measures, preventing production failures or a significant decline in yarn product quality due to excessive wear of the false twist component.

[0140] In this embodiment of the disclosure, when the fault type is the target fault, a prompt message is output.

[0141] When the fault type is determined to be the target fault, a prompt message is output. The prompt message can take various forms, such as audible and visual alarms, text prompts, SMS notifications, etc.

[0142] For example, if the target fault is a crack in the disc of the false twist component, the output prompt message could be: "A crack has been detected in the disc of the false twist component, accompanied by yarn slippage, severe tension fluctuations, and an abnormal increase in yarn hairiness; please stop the machine immediately, replace the worn disc, and then restart production."

[0143] For example, if the target fault is damage to the inner hole of the false twist component, the output prompt message could be: "Damage to the inner hole of the false twist component has been detected, accompanied by yarn scratches, unstable false twist effect, and a significant increase in yarn breakage rate; please stop the machine to repair the damaged inner hole area, or directly replace the false twist component body."

[0144] In this embodiment of the disclosure, when the fault type is the target fault, the corresponding prompt information is output, which can promptly and effectively convey the fault information to the operator so that appropriate handling measures can be taken to further ensure the quality and performance of the yarn after texturing.

[0145] Based on the same technical concept, this disclosure also provides a texturing control device 500 based on a false twist component, such as... Figure 5 As shown, it includes: The first input module 501 is used to input the video segment of the current time window in the video stream acquired from the false twist component into the visual encoder to obtain the initial visual features; The second input module 502 is used to input the parameter sequence of active control parameters and passive monitoring parameters of the texturing machine where the false twist component is located, which belongs to the current time window, into the text encoder to obtain text features; The third input module 503 is used to input the initial visual features and text features into the large language model to construct key-value pair information of at least one decoding layer of the large language model; Optimization module 504 is used to optimize the visual feature representation in key-value pair information based on the controller; The prediction module 505 is used to predict the detection results of the false twist component based on the optimized visual feature representation. The detection results include at least the degree of wear and the remaining life. Control module 506 is used to control the control parameters of the texturing process based on the detection results.

[0146] In some embodiments, the active control parameters include at least one of the following: false twist ratio, hot box temperature, and oil tanker speed; Passive monitoring parameters include at least one of the following: yarn tension, breakage rate, continuous operating time of the false twist component, bearing temperature of the false twist component, and historical operating records of the texturing machine.

[0147] In some embodiments, the optimization module includes: The first acquisition unit is used to acquire the input information of the target decoding layer for the current step. The input information includes the current output result of the large language model and the current key-value pair information of the target decoding layer. The prediction unit is used to predict the input information using a large language model and obtain the confidence level of the intermediate output result of the current step. The first determining unit is used to determine the distribution entropy of the current step based on the confidence level; The second determining unit is used to determine the exponential moving average entropy of the current step based on the distribution entropy and the exponential moving average entropy of the previous step. The third determining unit is used to determine the target loss based on the exponential moving average entropy of the current step; The fourth determining unit is used to determine the gradient of the target loss with respect to the controller parameters of the controller; The first update unit is used to update the controller parameters based on the gradient to obtain the controller parameter update results; The second update unit is used to update the visual feature representation in the current key-value pair information based on the controller parameter update result; The second acquisition unit is used to construct the input information of the next step based on the updated visual feature representation when it is determined that the termination condition is not met, take the next step as the current step, and trigger the first acquisition unit to return to execute the operation of acquiring the input information of the target decoding layer for the current step.

[0148] In some embodiments, an analysis module is also included, for: Obtain reference video sequences and reference parameter sequences for the false twist component under normal operating conditions; Principal component analysis is performed on the reference video sequence to obtain the initial parameters of the first preset channel set of the controller; Principal component analysis is performed on the reference parameter sequence to obtain the initial parameters of the second preset channel set of the controller.

[0149] In some embodiments, a first fixing module is further included, for: Based on the prediction results of the current time window, if it is determined that the remaining life of the false twist component predicted in multiple consecutive time windows is less than the life threshold, the fixed controller parameters will no longer change.

[0150] In some embodiments, the control module includes: The adjustment unit is used to adjust the false twist ratio and the temperature of the first hot box based on the wear degree and remaining life of the false twist component, when the false twist component is in a first state; wherein, the false twist ratio is negatively correlated with the wear degree; the temperature of the hot box is increased in specified steps, and the adjusted temperature of the hot box is lower than the upper temperature limit; The first state implicitly indicates that the wear of the false-twist component is lower than the target amount; The temperature of the first heating chamber is used to soften the yarn.

[0151] In some embodiments, after determining the gradient of the target loss with respect to the controller parameters, a pruning module is further included, for: Compare the gradient and the upper bound of the gradient; If the gradient is greater than the upper limit of the gradient, the gradient is clipped to the upper limit of the gradient.

[0152] In some embodiments, a second fixing module is further included, for: When the false twist component is in the second state, the false twist ratio and the target hot box temperature used in the texturing process when the yarn breakage rate is the lowest within the target time range are obtained and used as the target false twist ratio and the target hot box temperature, respectively. The texturing process is fixed to use the target false twist ratio and the target hot box temperature; The second state is used to implicitly indicate that the wear of the false twist component is greater than or equal to the target amount.

[0153] In some embodiments, a second state determination module is further included, for determining that the false-twist component enters a second state if at least one of the following conditions is met: Condition 1: The decrease in the exponential moving average entropy over n consecutive time windows is greater than the target magnitude, and the distribution entropy of the current step is lower than the target percentage of the preset peak value. Condition 2: The exponential moving average entropy of the current step is less than the target threshold determined based on empirical values, and the decrease in the exponential moving average entropy of the current step compared to the preset peak value is greater than the preset magnitude.

[0154] In some embodiments, the detection result may also include the fault type of the false twist component, and the device may also include a prompting module for outputting prompting information when the fault type is the target fault.

[0155] Figure 6 This is a structural block diagram of an electronic device according to an embodiment of the present disclosure. Figure 6 As shown, the electronic device includes a memory 610 and a processor 620. The memory 610 stores a computer program that can run on the processor 620. There can be one or more memories 610 and processors 620. The memory 610 can store one or more computer programs, which, when executed by the electronic device, cause the electronic device to perform the methods provided in the above-described method embodiments. The electronic device may also include a communication interface 630 for communicating with external devices and performing data exchange and transmission.

[0156] If the memory 610, processor 620, and communication interface 630 are implemented independently, they can be interconnected via a bus to communicate with each other. This bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. This bus can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 6 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0157] Optionally, in a specific implementation, if the memory 610, processor 620, and communication interface 630 are integrated on a single chip, then the memory 610, processor 620, and communication interface 630 can communicate with each other through an internal interface.

[0158] It should be understood that the aforementioned processor can be a Central Processing Unit (CPU), or other general-purpose processors, Digital Signal Processors (DSPs), Application Specific Integrated Circuits (ASICs), Field-Programmable Gate Arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. General-purpose processors can be microprocessors or any conventional processor. It is worth noting that the processor can be a processor supporting Advanced Reduced Instruction Set Machines (ARM) architecture.

[0159] Further, optionally, the aforementioned memory may include read-only memory and random access memory, and may also include non-volatile random access memory. The memory may be volatile or non-volatile, or may include both. 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), which serves as an external cache. Many forms of RAM are available by way of example, but not limitation. Examples include Static Random Access Memory (SRAM), Dynamic Random Access Memory (DRAM), Synchronous DRAM (SDRAM), Double Data Rate SDRAM (DDR SDRAM), Enhanced Synchronous DRAM (ESDRAM), Synchlink DRAM (SLDRAM), and Direct RAMBUS RAM (DR RAM).

[0160] In the description of the embodiments of this disclosure, references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this disclosure. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of those different embodiments or examples.

[0161] In the description of the embodiments disclosed herein, unless otherwise stated, " / " means "or". For example, A / B can mean A or B. The "and / or" in this document is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone.

[0162] In the description of embodiments of this disclosure, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of embodiments of this disclosure, unless otherwise stated, "a plurality of" means two or more.

[0163] The above description is merely an exemplary embodiment of this disclosure and is not intended to limit this disclosure. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this disclosure should be included within the protection scope of this disclosure.

Claims

1. A texturing control method based on a false twist component, comprising: The video segment of the current time window in the video stream acquired from the false twist component is input into the visual encoder to obtain the initial visual features; The parameter sequence of active control parameters and passive monitoring parameters of the texturing machine where the false twist component is located, belonging to the current time window, is input into the text encoder to obtain text features; The initial visual features and the text features are input into a large language model to construct key-value pair information for at least one decoding layer of the large language model; The controller optimizes the visual feature representation in the key-value pair information. Based on the optimized visual feature representation, the detection result of the false twist component is predicted, and the detection result includes at least the degree of wear and the remaining life. Based on the test results, control parameters for the texturing process are controlled.

2. The method according to claim 1, wherein, The active control parameters include at least one of the following: false twist ratio, hot box temperature, and oil tanker speed; The passive monitoring parameters include at least one of the following: yarn tension, breakage rate, continuous operating time of the false twist component, bearing temperature of the false twist component, and historical operating records of the texturing machine.

3. The method according to claim 1, wherein, The controller-based optimization of the visual feature representation in the key-value pair information includes: For each target decoding layer in the at least one decoding layer, iteratively perform the following operations: For the current step, obtain the input information of the target decoding layer, which includes the current output result of the large language model and the current key-value pair information of the target decoding layer; The large language model is used to predict the input information to obtain the confidence level of the intermediate output result of the current step; The distribution entropy of the current step is determined based on the confidence level; The exponential moving average entropy of the current step is determined based on the distribution entropy and the exponential moving average entropy of the previous step. The target loss is determined based on the exponential moving average entropy of the current step; Determine the gradient of the target loss with respect to the controller parameters of the controller; The controller parameters are updated based on the gradient to obtain the controller parameter update result; The visual feature representation in the current key-value pair information is updated based on the controller parameter update result; If the termination condition is not met, the input information for the next step of the current step is constructed based on the updated visual feature representation. The next step is then taken as the current step, and the operation of obtaining the input information of the target decoding layer for the current step is returned.

4. The method according to claim 3, further comprising: The initial values ​​of the controller parameters are determined based on the following method: Obtain a reference video sequence and a reference parameter sequence of the false twist component under normal operating conditions; Principal component analysis is performed on the reference video sequence to obtain the initial parameters of the first preset channel set of the controller; Principal component analysis is performed on the reference parameter sequence to obtain the initial parameters of the second preset channel set of the controller.

5. The method according to claim 3, further comprising: Based on the prediction results of the current time window, if it is determined that the remaining lifespan of the false twist component predicted in multiple consecutive time windows is less than the lifespan threshold, the controller parameters are fixed and no longer changed.

6. The method according to claim 1, wherein, The control parameters for the texturing process based on the detection results include: Based on the wear degree and remaining life of the false twist component, when the false twist component is in a first state, the false twist ratio and the temperature of the first hot box are adjusted based on the wear degree; wherein, the false twist ratio is negatively correlated with the wear degree; the temperature of the hot box is increased in specified steps, and the adjusted temperature of the hot box is lower than the upper temperature limit; The first state implicitly indicates that the wear of the false twist component is lower than the target amount; The temperature of the first heating box is used to soften the yarn.

7. The method of claim 6, further comprising, after determining the gradient of the target loss with respect to the controller parameters of the controller: Compare the gradient with the upper bound of the gradient; If the gradient is greater than the upper limit of the gradient, the gradient is clipped to the upper limit of the gradient.

8. The method according to claim 6, further comprising: When the false twist component is determined to be in the second state, the false twist ratio used in the texturing process and the target hot box temperature are obtained when the yarn breakage rate is the lowest within the target time range, and are respectively used as the target false twist ratio and the target hot box temperature. The texturing process is fixed to use the target false twist ratio and the target hot box temperature; The second state implicitly indicates that the wear of the false twist component is greater than or equal to the target amount.

9. The method according to claim 8, further comprising: The false-twist component is determined to enter the second state if at least one of the following conditions is met: Condition 1: The decrease in the exponential moving average entropy over n consecutive time windows is greater than the target magnitude, and the distribution entropy of the current step is lower than the target percentage of the preset peak value. Condition 2: The exponential moving average entropy of the current step is less than the target threshold determined based on empirical values, and the decrease of the exponential moving average entropy of the current step compared to the preset peak value is greater than the preset magnitude.

10. The method according to any one of claims 1-9, wherein the detection result further includes the fault type of the false twist component, and the method further includes: If the fault type is the target fault, a prompt message will be output.

11. A texturing control device based on a false twist component, comprising: The first input module is used to input the video segment of the current time window in the video stream acquired from the false twist component into the visual encoder to obtain the initial visual features; The second input module is used to input the parameter sequence of active control parameters and passive monitoring parameters of the texturing machine where the false twist component is located, which belongs to the current time window, into the text encoder to obtain text features; The third input module is used to input the initial visual features and the text features into the large language model to construct key-value pair information of at least one decoding layer of the large language model; The optimization module is used to optimize the visual feature representation in the key-value pair information based on the controller; A prediction module is used to predict the detection result of the false twist component based on the optimized visual feature representation, the detection result including at least the degree of wear and the remaining life; The control module is used to control the control parameters of the texturing process based on the detection results.

12. An electronic device, comprising: At least one processor; as well as A memory communicatively connected to the at least one processor; wherein, The memory stores instructions executable by the at least one processor, which, when executed by the at least one processor, enables the at least one processor to perform the method of any one of claims 1-10.

13. A non-transitory computer-readable storage medium storing computer instructions, wherein, The computer instructions are used to cause the computer to perform the method according to any one of claims 1-10.

14. A computer program product comprising a computer program that, when executed by a processor, implements the method according to any one of claims 1-10.

Images