Dynamic scheduling management method and system for clothing production

Abstract

The invention relates to the field of garment production, and particularly discloses a dynamic scheduling management method and system for garment production, and the method comprises the steps: firstly obtaining the production process data of a garment order, the production schedule data of the garment order and the position information of garment production equipment, and then carrying out the feature extraction and correlation analysis of the three through a deep learning technology, and finally, a garment manufacturing order production process scheduling table is generated through a generator, so that the production processes of different garments with the same production process data are arranged on the same equipment, the accuracy and efficiency of production scheduling are improved, resource waste is reduced, and the production flexibility is improved.

Description

Technical Field

[0001] The present application relates to the field of garment production, and more specifically, to a dynamic scheduling management method and system for garment production. Background Art

[0002] Garment making refers to the process of transforming textile raw materials into finished garments through cutting, sewing, decoration and other crafts. The garment industry is not only the inheritance of traditional handicrafts, but also an important field of modern industrialized production. Its products are widely used in daily life, fashion trends, special functional needs and other fields. The garment industry gathers a large number of labor resources, and its product orders tend to be personalized, with a variety of styles and changing at any time.

[0003] At present, most garment manufacturing companies still rely on manual experience to schedule garment production, with a relatively low level of intelligence. They are unable to fully tap the potential of workshop manufacturing resources, resulting in low utilization of workshop equipment and slow response to orders, affecting the punctuality and agility of production.

[0004] Therefore, a dynamic scheduling management method and system for garment production is desired. Summary of the invention

[0005] In order to solve the above technical problems, this application is proposed. The embodiment of this application provides a dynamic scheduling management method and system for garment production, which first obtains the production process data of the garment order, the production scheduling data of the garment order and the location information of the garment production equipment, and then uses deep learning technology to perform feature extraction and association analysis on the three, and finally generates a garment order production process scheduling table through a generator, so that the production processes of different garments with the same production process data are arranged on the same equipment, thereby improving the accuracy and efficiency of production scheduling, thereby reducing resource waste and improving production flexibility.

[0006] According to one aspect of the present application, a dynamic scheduling management method for garment production is provided, which includes: Obtain production process data, production schedule data and location information of garment production equipment for garment orders; Extracting a garment order multimodal information association feature vector and a garment production equipment location information semantic feature vector from the production process data of the garment order, the production scheduling data of the garment order and the garment production equipment location information; A garment order production process scheduling table is generated based on the garment order multimodal information association feature vector and the garment production equipment location information semantic feature vector.

[0007] According to another aspect of the present application, a dynamic scheduling management system for garment production is provided, which includes: The clothing production information acquisition module is used to obtain the production process data of clothing orders, the production scheduling data of clothing orders and the location information of clothing production equipment; A clothing production information extraction module, used to extract a clothing order multimodal information association feature vector and a clothing production equipment location information semantic feature vector from the production process data of the clothing order, the production scheduling data of the clothing order and the clothing production equipment location information; A clothing order production process scheduling table generation module is used to generate a clothing order production process scheduling table based on the clothing order multimodal information association feature vector and the clothing production equipment location information semantic feature vector.

[0008] Compared with the prior art, the present application provides a dynamic scheduling management method and system for garment production, which first obtains the production process data, production scheduling data and location information of garment production equipment of a garment order, and then uses deep learning technology to perform feature extraction and correlation analysis on the three, and finally generates a production process scheduling table for a garment order through a generator, so as to arrange the production processes of different garments with the same production process data on the same equipment, thereby improving the accuracy and efficiency of production scheduling, thereby reducing resource waste and improving production flexibility. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0010] Figure 1 Flow chart of a dynamic scheduling management method for garment production according to an embodiment of the present application.

[0011] Figure 2 The present invention is a flowchart of extracting features from the production process data of a clothing order in a dynamic scheduling management method for clothing production according to an embodiment of the present application to obtain a clothing order production process-related feature vector.

[0012] Figure 3 The present invention is a flowchart of extracting features from the production scheduling data of a clothing order in a dynamic scheduling management method for clothing production according to an embodiment of the present application to obtain a feature vector of the clothing order production scheduling text.

[0013] Figure 4The present invention is a flowchart of extracting features from the clothing production equipment location information in the dynamic scheduling management method for clothing production according to an embodiment of the present application to obtain a semantic feature vector of the clothing production equipment location information.

[0014] Figure 5 It is a block diagram of a dynamic scheduling management system for garment production according to an embodiment of the present application. DETAILED DESCRIPTION

[0015] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.

[0016] Figure 1 FIG. 1 is a flow chart of a dynamic scheduling management method for garment production according to an embodiment of the present application. Figure 1 As shown, the dynamic scheduling management method for garment production according to the embodiment of the present application includes: S110, acquiring the production process data of the garment order, the production scheduling data of the garment order and the location information of the garment production equipment; S120, extracting the garment order multimodal information association feature vector and the garment production equipment location information semantic feature vector from the production process data of the garment order, the production scheduling data of the garment order and the garment production equipment location information; S130, generating a garment order production process scheduling table based on the garment order multimodal information association feature vector and the garment production equipment location information semantic feature vector.

[0017] In the above-mentioned dynamic scheduling management method for garment production, the step S110 obtains the production process data of the garment order, the production scheduling data of the garment order and the location information of the garment production equipment. It should be understood that garment making is the process of transforming textile materials into finished garments through processes such as cutting, sewing and decoration. This industry not only continues the essence of traditional handicrafts, but is also an important part of modern industrialized production. Its products are widely used in many fields such as daily life, fashion trends and specific functional needs. The garment industry gathers a large number of laborers, and orders often require personalized customization, with diverse and constantly changing styles. However, most garment companies currently still rely mainly on manual experience for production scheduling, with a low degree of intelligence, and fail to fully utilize the potential of workshop manufacturing resources. This leads to insufficient utilization of workshop equipment and slow order response speed, which affects the punctuality and flexibility of production. Therefore, in the technical solution of the present application, the production process data of the garment order, the production scheduling data of the garment order and the location information of the garment production equipment are obtained, and combined with deep learning technology, the production process scheduling table of the garment order is generated to arrange the production steps of different garments with the same production process on the same equipment. This approach aims to improve the accuracy and efficiency of production scheduling, thereby reducing waste of resources and increasing production flexibility.

[0018] Specifically, production process data includes detailed information such as the material, size, sewing method, and process sequence of each product. With accurate process data, enterprises can effectively carry out process design and standardized production. Production scheduling data involves information such as the delivery time, production sequence, and working hours of each order. Through effective scheduling management, enterprises can avoid waste of resources in the production process and ensure on-time delivery. The location information of clothing production equipment is related to the layout and efficiency of the production line. By understanding the location information of the equipment, enterprises can reasonably plan the production line, improve production fluency, and reduce material handling time. Obtaining the production process data, production scheduling data, and clothing production equipment location information of clothing orders will not only help improve production efficiency and resource utilization, but also ensure product quality and delivery timeliness. The effective integration and management of these three will drive clothing companies to achieve greater success in the fierce market competition.

[0019] In the above-mentioned dynamic scheduling management method for garment production, the step S120 extracts the garment order multimodal information association feature vector and the garment production equipment location information semantic feature vector from the garment order production process data, the garment order production scheduling data and the garment production equipment location information. It should be understood that extracting the multimodal information association feature vector and the semantic feature vector from the garment order production process data, the production scheduling data and the equipment location information enhances the utilization value of the data and provides a more accurate decision-making basis for the enterprise. In this way, the enterprise can optimize production, control costs and predict delivery time based on these feature vectors, and ultimately improve the overall market competitiveness and customer satisfaction.

[0020] In a specific embodiment of the present application, the step S120 includes: performing feature extraction on the production process data of the garment order to obtain a garment order production process associated feature vector; performing feature extraction on the production scheduling data of the garment order to obtain a garment order production scheduling text feature vector; associating the garment order production process associated feature vector and the garment order production scheduling text feature vector to obtain the garment order multimodal information associated feature vector; performing feature extraction on the clothing production equipment location information to obtain the clothing production equipment location information semantic feature vector.

[0021] It should be understood that production process data usually contains a variety of information, such as fabric type, sewing method, process sequence, time estimation, etc. This information is often multidimensional and complex, and directly using raw data for analysis may lead to noise interference and information redundancy. Through feature extraction, key variables and patterns can be identified to form a more concise and meaningful feature vector. This vector can effectively capture the core characteristics of the production process and facilitate subsequent analysis and decision-making.

[0022] Furthermore, production scheduling data usually includes information from multiple aspects, such as order number, delivery time, production priority, process arrangement and resource allocation. This information is often described in text form and contains rich production and scheduling information. Directly processing these raw text data may lead to information redundancy and noise, affecting the subsequent analysis effect. Through feature extraction, the raw text can be converted into meaningful feature vectors, and the key production scheduling features can be extracted to make the data more concise and easy to analyze. By analyzing the feature vectors, enterprises can identify the key factors affecting production scheduling. For example, some orders may need to adjust production priorities due to insufficient fabric supply or equipment failure. Feature extraction can not only help identify these problems, but also provide decision-making basis for production scheduling to ensure on-time delivery and customer satisfaction.

[0023] Furthermore, the production process feature vector mainly describes the specific process, materials and technical requirements of the production process, while the production scheduling text feature vector provides information about order priority, delivery time and production arrangement. These two types of features are complementary in function and content, providing key data from the perspectives of production execution and scheduling management respectively. By linking the two, enterprises can form a comprehensive perspective and gain a deep understanding of each link in the production process.

[0024] In particular, equipment location information is crucial data in the production process. It not only includes the physical location of the equipment, but also involves the function, working status and relationship of the equipment with other equipment. Directly using the original location information for analysis may lead to data redundancy and information confusion. Through feature extraction, key semantic features can be extracted from complex location information, making the data more concise and easy to analyze. Such semantic feature vectors can effectively express the function and status of the equipment, providing strong support for subsequent data analysis and decision-making.

[0025] Figure 2 The present invention is a flowchart of extracting features from the production process data of a clothing order to obtain a clothing order production process-related feature vector in the dynamic scheduling management method for clothing production according to an embodiment of the present application. Figure 2 As shown, in a specific embodiment of the present application, feature extraction is performed on the production process data of the garment order to obtain a garment order production process associated feature vector, including: S210, embedding the production process data of the garment order into an encoder through a garment order production process to obtain a plurality of garment order production process embedding vectors; S220, passing the plurality of garment order production process embedding vectors through a garment order production process feature extractor based on a deep neural network to obtain a plurality of garment order production process feature vectors; S230, cascading the plurality of garment order production process feature vectors to obtain the garment order production process associated feature vector.

[0026] It should be understood that production process data usually contains rich dimensional information, such as fabric type, process arrangement, technical requirements, and equipment requirements. Although this information is of great value, it is difficult to directly analyze effectively in its original state. Using an embedded encoder, complex production process data can be mapped to a low-dimensional vector space, making the data easier to process and understand. The embedded vector can effectively capture the potential characteristics and relationships of the production process, making subsequent analysis and machine learning tasks more efficient. Through the encoder, the relationship between the various features of the production process data is more clearly presented. Different process features can form similar vector representations in the embedded space, so that orders with similar production processes can be closer in the vector space. Such feature correlation can help companies quickly identify similar orders, thereby achieving more efficient management in production scheduling and resource allocation. For example, companies can match new orders for similarity based on the production process vectors of historical orders to better plan production resources.

[0027] Furthermore, the embedding vector has transformed the original production process data into an effective low-dimensional representation, but to further refine the features that are more targeted for production decisions, it is necessary to perform deeper feature extraction through deep neural networks. Deep neural networks have powerful nonlinear modeling capabilities and can capture implicit features and relationships in complex data. This ability enables deep neural networks to learn higher-level semantic information based on the embedding vectors, providing richer and more useful features for subsequent analysis. Through the feature extractor based on deep neural networks, the production process embedding vectors of multiple garment orders can be further processed to generate more abstract and efficient feature vectors. These feature vectors not only retain the original process information, but also combine the potential relationships and patterns learned by the deep network to form more representative production process features. Specifically, each layer of the garment order production process feature extractor based on the deep neural network is used to perform convolution processing, mean pooling processing based on the local feature matrix and nonlinear activation processing on the input data in the forward pass of the layer, so that the last layer of the garment order production process feature extractor based on the deep neural network outputs the multiple garment order production process feature vectors, wherein the input of the garment order production process feature extractor based on the deep neural network is the multiple garment order production process embedding vectors.

[0028] Furthermore, in the garment production process, the production process feature vector of each order often only captures information of specific dimensions, such as fabric type, process arrangement, and production efficiency. However, the production process is often multidimensional and complex, and it is difficult to fully reflect the relevance and integrity of the production process by relying on a single feature vector. By cascading different production process feature vectors, the generated associated feature vectors can provide more contextual information and help the model learn more complex patterns and relationships. For example, when the process features of a production order are input into the model together with the features of other orders, the model can capture the similarities and differences between these orders, thereby optimizing production scheduling and resource allocation.

[0029] Figure 3 The present invention is a flowchart of extracting features from the production scheduling data of a clothing order in the dynamic scheduling management method for clothing production according to an embodiment of the present application to obtain a feature vector of the clothing order production scheduling text. Figure 3 As shown, in a specific embodiment of the present application, feature extraction is performed on the production scheduling data of the clothing order to obtain a clothing order production scheduling text feature vector, including: S310, word vector conversion is performed on the production scheduling data of the clothing order to obtain multiple clothing order production scheduling text embedding vectors; S320, the multiple clothing order production scheduling text embedding vectors are passed through a clothing order production scheduling text feature encoder based on multiple convolutional layers to obtain the clothing order production scheduling text feature vector.

[0030] It should be understood that in the garment production process, the scheduling texts of different orders may have similar structures and meanings. Through word vector conversion, the semantic relationship between these texts can be captured. For example, the production scheduling of some orders may involve similar processes or time arrangements, and word vectors can reflect these similarities in the vector space. The extraction of this semantic information helps the model to more effectively identify similar orders when performing cluster analysis and classification tasks, thereby optimizing production scheduling and resource allocation.

[0031] Furthermore, convolutional neural networks (CNNs) initially achieved remarkable success in the field of image processing, but their application in text data processing is equally effective. Through convolution operations, CNNs can automatically learn local features in text, such as word combinations and the meaning of phrases. This feature learning ability enables the model to capture the complex patterns and structures contained in production scheduling text. Among them, through multiple convolutional layers, the model can extract features at different levels. Each convolutional layer can be regarded as a deeper abstraction and processing of the input data. Lower convolutional layers can capture basic word features, while higher convolutional layers can capture more complex semantic relationships and contextual information. This hierarchical feature extraction method can make the final text feature vector more comprehensive and expressive, thereby improving the performance of the model in downstream tasks. For example, in tasks such as predicting production efficiency and evaluating order priority, the richness of feature vectors can significantly improve the accuracy and robustness of the model. Specifically, the clothing order production scheduling text feature encoder based on multiple convolutional layers includes: a first convolutional layer, a second convolutional layer parallel to the first convolutional layer, and a cascade layer connected to the first convolutional layer and the second convolutional layer, wherein the first convolutional layer uses a one-dimensional convolutional kernel with a first scale, and the second convolutional layer uses a one-dimensional convolutional kernel with a second scale. More specifically, the first convolutional layer of the clothing order production scheduling text feature encoder based on multiple convolutional layers is used to perform one-dimensional convolution encoding on the input data to obtain a first-scale feature vector; the second convolutional layer of the clothing order production scheduling text feature encoder based on multiple convolutional layers is used to perform one-dimensional convolution encoding on the input data to obtain a second-scale feature vector; the first-scale feature vector and the second-scale feature vector are cascaded to obtain the clothing order production scheduling text feature vector.

[0032] In a specific embodiment of the present application, the garment order production process associated feature vector and the garment order production scheduling text feature vector are associated to obtain the garment order multimodal information associated feature vector, including: installing a Java development environment JDK and necessary dependent libraries for vector calculation; creating a ServerSocket instance named MultiModalFeatureServer, listening to a specified port, and waiting for a client connection; when the client connects to the server, ServerSocket returns a Socket object, through which the client communicates; receiving the garment order production process associated feature vector and the garment order production scheduling text feature vector from the client and parsing them; using Apache CommonsMath to calculate the product of the two feature vectors, and sending the calculated garment order multimodal information associated feature vector back to the client; after the calculation is completed, closing the Socket connection and releasing resources.

[0033] It should be understood that ServerSocket is a class in Java used to implement server-side network communication. It allows server applications to listen to specific ports, wait for client connections and communicate. The main advantages of ServerSocket are its ease of use, strong reliability and high scalability. Through ServerSocket, the server can handle multiple client connections at the same time, ensuring the efficiency and stability of concurrent requests. In addition, the ServerSocket mechanism provides a rich API and error handling mechanism, making it easy to implement complex network communication functions.

[0034] The server-side deployment code is shown below.

[0035] import java.io.*; import java.net.ServerSocket; import java.net.Socket; import org.apache.commons.math3.linear.ArrayRealVector; import org.apache.commons.math3.linear.RealVector; public class MultiModalFeatureServer { public static void main(String[] args) { int port = 9090; / / Select an unused port try (ServerSocket serverSocket = new ServerSocket(port)) { System.out.println("Server is listening on port " +port); while (true) { Socket socket = serverSocket.accept(); System.out.println("New client connected"); / / Create input and output streams BufferedReader in = new BufferedReader(newInputStreamReader(socket.getInputStream())); PrintWriter out = new PrintWriter(socket.getOutputStream(), true); / / Read the feature vector sent by the client String featureVector1Str = in.readLine(); String featureVector2Str = in.readLine(); / / parse the feature vector RealVector featureVector1 = parseVector(featureVector1Str); RealVector featureVector2 = parseVector(featureVector2Str); / / Calculate the product RealVector resultVector = featureVector1.ebeMultiply(featureVector2); / / Send the result back to the client out.println(resultVector.toString()); / / Close the connection socket.close(); System.out.println("Client disconnected"); } } catch (IOException ex) { System.out.println("Server exception: " + ex.getMessage()); ex.printStackTrace(); } } private static RealVector parseVector(String vectorStr) { String[] elements = vectorStr.split(","); double[] data = new double[elements.length]; for (int i = 0; i <elements.length; i++) { data[i] = Double.parseDouble(elements[i]); } return new ArrayRealVector(data); } } The client deployment code is as follows.

[0036] import java.io.*; import java.net.Socket; public class MultiModalFeatureClient { public static void main(String[] args) { String hostname = "localhost"; int port = 9090; try (Socket socket = new Socket(hostname, port)) { / / Create input and output streams PrintWriter out = new PrintWriter(socket.getOutputStream(), true); BufferedReader in = new BufferedReader(newInputStreamReader(socket.getInputStream())); / / Send feature vector out.println("Clothing order production process associated feature vector"); / / Production process associated feature vector out.println("Garment order production scheduling text feature vector"); / / Production scheduling text feature vector / / Read the result returned by the server String result = in.readLine(); System.out.println("Result: " + result); } catch (IOException ex) { System.out.println("Client exception: " + ex.getMessage()); ex.printStackTrace(); } } } Here, the calculation service of multimodal information associated feature vector of clothing order based on ServerSocket mechanism is realized. This service mainly consists of server (MultiModalFeatureServer) and client (MultiModalFeatureClient).

[0037] On the server side, first, make sure that the Java Development Environment (JDK) and Apache CommonsMath library are installed on the server. Then, the server-side program creates a ServerSocket instance, listens to the specified port (such as 9090), and waits for the client to connect. Next, when the client connects, the server creates a Socket object to communicate with the client. Further, the server receives two feature vectors (production process association feature vector and production scheduling text feature vector) from the client and parses them into RealVector objects in the Apache Commons Math library. Furthermore, the ebeMultiply method is used to calculate the element-by-element product of the two feature vectors to obtain the multimodal information association feature vector. Finally, the calculation result is sent back to the client, the Socket connection is closed, and the resources are released.

[0038] In the client, first, the client connects to the host name and port specified by the server. Then, the client sends two feature vectors (production process association feature vector and production schedule text feature vector) to the server. Then, the client receives the multimodal information association feature vector returned by the server.

[0039] In this way, an efficient and reliable clothing order multimodal information correlation feature vector calculation service is implemented. Through the ServerSocket mechanism, the server can handle multiple client requests at the same time, ensuring the concurrency and scalability of the calculation task. The use of the Apache Commons Math library simplifies the implementation of vector calculation and improves the readability and maintainability of the code.

[0040] The server can quickly parse the feature vector sent by the client and use efficient vector calculation methods (such as ebeMultiply) to calculate the multimodal information correlation feature vector. The client can send feature vectors and receive calculation results in real time, which is suitable for application scenarios that require fast response. The ServerSocket mechanism allows the server to handle multiple client connections at the same time, which is suitable for scenarios with large-scale concurrent requests. The code structure is clear, and the dependent libraries (such as Apache CommonsMath) provide rich mathematical calculation functions, simplifying development and maintenance.

[0041] Figure 4 This is a flow chart of extracting features of the clothing production equipment location information to obtain the semantic feature vector of the clothing production equipment location information in the dynamic scheduling management method for clothing production according to an embodiment of the present application. Figure 4 As shown, in a specific embodiment of the present application, feature extraction is performed on the clothing production equipment location information to obtain a semantic feature vector of the clothing production equipment location information, including: S410, segmenting the clothing production equipment location information to obtain a clothing production equipment location information word sequence; S420, passing the clothing production equipment location information word sequence through a clothing production equipment location information semantic encoder based on a context encoder to obtain the clothing production equipment location information semantic feature vector.

[0042] It should be understood that in clothing production, equipment location information may contain various forms of descriptions, such as specific production workshops, machine numbers, relative positions, etc. This information is recorded in text form and is difficult to be directly processed by computers. Through word segmentation, this natural language text can be split into words or phrases with independent meanings to form an ordered word sequence. The word sequence after word segmentation can better represent the various components of equipment location information. Among them, the location information of clothing production equipment often involves multiple levels of information, including production lines, equipment types, location markers, etc. Through word segmentation, these different dimensions of information can be separated from the text. This provides support for the production management system to perform accurate equipment positioning, status monitoring, and path planning.

[0043] Furthermore, in clothing production, equipment location information often exists in the form of word sequences. These words are not only independent information, but also form semantic associations with surrounding words. Context-based encoders (such as long short-term memory networks LSTM or Transformer architectures) can effectively capture this contextual information, allowing the model to understand the relationship between different devices and their role in the production process when processing location information. Considering that traditional methods often rely only on statistical features such as bag-of-words models or TF-IDF, these methods cannot effectively capture semantic information in text. By using context encoders, the model can learn deep semantic features of equipment location information. This encoding not only takes into account the literal meaning of words, but also understands the meaning of words in a specific context, thereby improving the richness and expressiveness of the final semantic feature vector. Specifically, the embedding layer of the context encoder-based clothing production equipment position information semantic encoder is used to convert the clothing production equipment position information word sequence into an embedding vector to obtain a sequence of position information word embedding vectors; the converter-based Bert model of the context encoder-based clothing production equipment position information semantic encoder is used to perform global context semantic encoding on the sequence of position information word embedding vectors to obtain multiple position information feature vectors; and, the multiple position information feature vectors are cascaded to obtain the clothing production equipment position information semantic feature vector.

[0044] In the above-mentioned dynamic scheduling management method for garment production, the step S130 generates a garment order production process scheduling table based on the garment order multimodal information association feature vector and the garment production equipment location information semantic feature vector. It should be understood that the multimodal information of the garment order contains a rich production background. This information not only includes traditional data such as the quantity, style and delivery time of the order, but may also involve the personalized needs of customers, historical order data, market trend analysis, etc. These diverse data sources enable a comprehensive understanding of the characteristics and requirements of each order. The semantic feature vector of the equipment location information provides the real-time status of the production environment. This information includes the current availability of the equipment, workload, maintenance status and layout of the production line. By integrating multimodal information with equipment location information, a comprehensive production scheduling model can be generated, which takes into account the order characteristics and equipment status in real time. This comprehensive consideration allows the scheduling table to be quickly adjusted in a changing environment.

[0045] In a specific embodiment of the present application, the step S130 includes: fusing the garment order multimodal information association feature vector and the garment production equipment location information semantic feature vector to obtain a garment order scheduling generation feature vector; performing feature modulation optimization based on learning parameter structure redundancy on the garment order scheduling generation feature vector to obtain a sparse optimized garment order scheduling generation feature vector; passing the sparse optimized garment order scheduling generation feature vector through a generator to generate a garment order production process scheduling table.

[0046] It should be understood that by analyzing order characteristics and equipment status, the fused feature vector can provide decision support for production scheduling. For example, the system can automatically identify which equipment is most suitable for processing specific orders, which orders can be processed first, and which orders can be combined to reduce standby time, thereby reducing delivery time. Such optimization not only improves production efficiency, but also reduces production costs and enhances the market competitiveness of enterprises.

[0047] In particular, in the technical solution of the present application, the feature vector generated by the scheduling of clothing orders is obtained by fusing multiple features (such as production process features, production scheduling features, and equipment location information features). These features not only have different dimensions and information content, but may also overlap and intersect in semantics. For example, production process features and production scheduling features may reflect the same production efficiency or time requirements in some cases, but are expressed in different forms, resulting in information duplication. When these feature vectors are fused, if the fusion strategy is inappropriate, the final feature vector generated by the scheduling of clothing orders may contain redundant information. This redundancy not only increases the complexity of the calculation, but may also cause the model to learn redundant and irrelevant features, resulting in interference and confusion between features. Furthermore, if the dimension of the learning parameter matrix is ​​too large, it may lead to too many degrees of freedom, so that the model can overfit the noise in the training data; if the dimension is too small, it may not be able to fully learn important features, thereby affecting the prediction performance of the model, resulting in overfitting of the model during the learning process, and affecting the quality of the final generated schedule. Therefore, in the technical solution of the present application, the clothing order scheduling generation feature vector is subjected to feature modulation optimization based on the learning parameter structure redundancy to obtain a sparse optimized clothing order scheduling generation feature vector.

[0048] Among them, the feature modulation optimization based on the learning parameter structure redundancy is performed on the clothing order scheduling generation feature vector to obtain the sparse optimized clothing order scheduling generation feature vector, including: extracting each row vector of the learning parameter matrix in the generator as a set of clothing order scheduling learning parameter row vectors; calculating the clothing order scheduling entropy degradation maximization simulation value between the clothing order scheduling generation feature vector and each clothing order scheduling learning parameter row vector in the set of clothing order scheduling learning parameter row vectors to obtain a set of clothing order scheduling entropy degradation maximization simulation values; based on the comparison between each clothing order scheduling entropy degradation maximization simulation value in the set of clothing order scheduling entropy degradation maximization simulation values ​​and a preset threshold, the learning parameter matrix is ​​gated and pruned to obtain the clothing order scheduling learning parameter entropy degradation optimization matrix; the clothing order scheduling generation feature vector is matrix-multiplied with the clothing order scheduling learning parameter entropy degradation optimization matrix to obtain the sparse optimized clothing order scheduling generation feature vector.

[0049] The optimization steps are specifically expressed as follows: ;in, The set of row vectors of learning parameters for garment order scheduling, is a set operation, , , , are the first, second, and third row vectors of the clothing order scheduling learning parameters. , The eigenvalues ​​of the row vector of learning parameters for clothing order scheduling, The set of values ​​for maximizing the entropy degradation of clothing order scheduling is The entropy degradation maximization simulation value of the clothing order scheduling is is the maximum value function, Represents the calculation vector and The Euclidean distance between Generate feature vectors for clothing order scheduling For the The transposed vector of the row vector of clothing order scheduling learning parameters, is the length of the feature vector, is matrix multiplication, , , , The first, second, and , learning parameter mask values, For mask operation, Indicates the preset threshold value, Learning parameter entropy degradation optimization matrix for garment order scheduling, Generating feature vectors for sparse optimal clothing order scheduling.

[0050] In the technical solution of the present application, it is taken into consideration that in the process of inputting the garment order scheduling generating feature vector into the learning parameter matrix of the generator to obtain the garment order production process scheduling table, there is a mismatch between the structure of the learning parameter matrix itself and the garment order scheduling generating feature vector, that is, the learning parameter matrix has structural redundancy and information confusion relative to the garment order scheduling generating feature vector, which may cause feature redundancy or noise in the garment order production process scheduling table, resulting in model overfitting.

[0051] Based on this, in the technical solution of the present application, the clothing order scheduling generating feature vector is optimized by feature modulation based on the redundancy of the learning parameter structure, which uses the clothing order scheduling generating feature vector as a wandering node, and uses the feature distribution between the clothing order scheduling generating feature vector and each row vector in the learning parameter matrix to simulate the wandering mode, and uses the distance topology between the clothing order scheduling generating feature vector and each row vector in the learning parameter matrix as a wandering map. In this way, the walking result of the wandering node in a specific wandering mode on the wandering map is used to obtain the clothing order scheduling entropy degradation maximization simulation value. Furthermore, based on the comparison between each clothing order scheduling entropy degradation maximization simulation value in the set of clothing order scheduling entropy degradation maximization simulation values ​​and a preset threshold, the part of the learning parameter matrix that does not contribute significantly to the input data representation and may even introduce noise or redundant information is screened out. Specifically, based on the comparison result, the learning parameter matrix is ​​gated and pruned to obtain the clothing order scheduling learning parameter entropy degradation optimization matrix. Finally, the structurally optimized learning parameter matrix is ​​used as a feature modulation unit to linearly modulate the clothing order scheduling generation feature vector to obtain the sparse optimized clothing order scheduling generation feature vector.

[0052] Furthermore, traditional scheduling methods often rely on manual experience, which may be subjective and uncertain. However, through the combination of sparse optimization of clothing order scheduling to generate feature vectors and generators, enterprises can use data-driven methods to formulate scheduling strategies. By analyzing historical data and current order information, the generator can generate a more scientific production process schedule, effectively reducing the risk of scheduling decisions.

[0053] In summary, the embodiment of the present application first obtains the production process data of the garment order, the production scheduling data of the garment order and the location information of the garment production equipment, and then uses deep learning technology to perform feature extraction and correlation analysis on the three, and finally uses a generator to generate a garment order production process scheduling table, so as to arrange the production processes of different garments with the same production process data on the same equipment, thereby improving the accuracy and efficiency of production scheduling, thereby reducing resource waste and improving production flexibility.

[0054] Figure 5 FIG. 1 is a block diagram of a dynamic scheduling management system for garment production according to an embodiment of the present application. Figure 5 As shown, according to the dynamic scheduling management system 100 of garment production according to the embodiment of the present application, it includes: a garment production information acquisition module 110, which is used to acquire the production process data of a garment order, the production scheduling data of a garment order and the location information of garment production equipment; a garment production information extraction module 120, which is used to extract the garment order multimodal information association feature vector and the garment production equipment location information semantic feature vector from the production process data of the garment order, the production scheduling data of the garment order and the location information of the garment production equipment; a garment order production process scheduling table generation module 130, which is used to generate a garment order production process scheduling table based on the garment order multimodal information association feature vector and the garment production equipment location information semantic feature vector.

[0055] Here, those skilled in the art will appreciate that the specific operations of each step in the above-mentioned dynamic scheduling management system for garment production have been described in detail above. Figures 1 to 4 The description of the dynamic scheduling management method for garment production has been introduced in detail, and therefore, its repeated description will be omitted.

[0056] As described above, the dynamic scheduling management system 100 for garment production according to the embodiment of the present application can be implemented in various terminal devices. In one example, the dynamic scheduling management system 100 for garment production can be integrated into the terminal device as a software module and / or a hardware module. For example, the dynamic scheduling management system 100 for garment production can be a software module in the operating system of the terminal device, or can be an application developed for the terminal device; of course, the dynamic scheduling management system 100 for garment production can also be one of the many hardware modules of the terminal device.

[0057] Alternatively, in another example, the dynamic scheduling management system 100 for garment production and the terminal device may also be separate devices, and the dynamic scheduling management system 100 for garment production may be connected to the terminal device via a wired and / or wireless network and transmit interactive information in accordance with an agreed data format.

[0058] The various methods / implementations described in this specification may be used alone or in combination, and may be switched during execution. In addition, the processing steps, sequences, flow charts, etc. of the various methods / implementations described in this specification may be changed in order as long as there is no contradiction. For example, with respect to the methods described in this specification, various step units are given in an exemplary order, and are not limited to the specific order given.

[0059] The phrase "based on" used in this specification does not mean "only based on" unless otherwise specified in other paragraphs. In other words, the phrase "based on" means both "only based on" and "at least based on".

[0060] Any reference to a unit using the names "first", "second", etc. used in this specification does not fully limit the number or order of these units. These names may be used in this specification as a convenient method to distinguish between two or more units. Therefore, a reference to a first unit and a second unit does not mean that only two units can be used or that the first unit must precede the second unit in some form.

[0061] When the terms "including", "comprising", and their variations are used in this specification or claims, these terms are open ended like the term "having". Furthermore, the term "or" used in this specification or claims is not an exclusive or.

[0062] Those skilled in the art will appreciate that various aspects of the present application may be illustrated and described by a number of patentable categories or situations, including any new and useful process, machine, product or combination of substances, or any new and useful improvements thereto. Accordingly, various aspects of the present application may be performed entirely by hardware, entirely by software (including firmware, resident software, microcode, etc.), or by a combination of hardware and software. The above hardware or software may all be referred to as "data blocks", "modules", "engines", "units", "components" or "systems". In addition, various aspects of the present application may be represented as a computer product located in one or more computer-readable media, which includes computer-readable program code.

[0063] Those skilled in the art will appreciate that various aspects of the present application may be illustrated and described by a number of patentable categories or situations, including any new and useful process, machine, product or combination of substances, or any new and useful improvements thereto. Accordingly, various aspects of the present application may be performed entirely by hardware, entirely by software (including firmware, resident software, microcode, etc.), or by a combination of hardware and software. The above hardware or software may all be referred to as "data blocks", "modules", "engines", "units", "components" or "systems". In addition, various aspects of the present application may be represented as a computer product located in one or more computer-readable media, which includes computer-readable program code.

[0064] Unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by one of ordinary skill in the art to which the present disclosure belongs. It should also be understood that terms such as those defined in common dictionaries should be interpreted as having a meaning consistent with their meaning in the context of the relevant technology and should not be interpreted in an idealized or highly formal sense, unless explicitly defined as such herein.

[0065] The present invention has been described in detail above, but it is obvious to those skilled in the art that the present invention is not limited to the embodiments described in this specification. The present invention can be implemented as a modification and variation without departing from the purpose and scope of the present invention as determined by the description of the claims. Therefore, the description of this specification is for the purpose of illustration and does not have any restrictive meaning for the present invention.

Claims

1. A dynamic scheduling management method for garment production, characterized in that: include: Obtain production process data, production schedule data and location information of garment production equipment for garment orders; Extracting a garment order multimodal information association feature vector and a garment production equipment location information semantic feature vector from the production process data of the garment order, the production scheduling data of the garment order and the garment production equipment location information; A garment order production process scheduling table is generated based on the garment order multimodal information association feature vector and the garment production equipment location information semantic feature vector.

2. The dynamic scheduling management method for garment production according to claim 1, characterized in that: Extracting a garment order multimodal information association feature vector and a garment production equipment location information semantic feature vector from the garment order production process data, the garment order production schedule data and the garment production equipment location information, including: Extracting features from the production process data of the clothing order to obtain a clothing order production process-related feature vector; Performing feature extraction on the production scheduling data of the clothing order to obtain a clothing order production scheduling text feature vector; Associating the clothing order production process associated feature vector with the clothing order production scheduling text feature vector to obtain the clothing order multimodal information associated feature vector; Feature extraction is performed on the clothing production equipment location information to obtain a semantic feature vector of the clothing production equipment location information.

3. The dynamic scheduling management method for garment production according to claim 2, characterized in that: Extracting features from the production process data of the clothing order to obtain a clothing order production process-related feature vector includes: Embedding the production process data of the clothing order into a clothing order production process encoder to obtain a plurality of clothing order production process embedding vectors; Passing the multiple garment order production process embedding vectors through a garment order production process feature extractor based on a deep neural network to obtain multiple garment order production process feature vectors; The multiple clothing order production process feature vectors are cascaded to obtain the clothing order production process associated feature vector.

4. The dynamic scheduling management method for garment production according to claim 3 is characterized in that: Performing feature extraction on the production scheduling data of the clothing order to obtain a clothing order production scheduling text feature vector includes: Performing word vector conversion on the production scheduling data of the clothing order to obtain multiple clothing order production scheduling text embedding vectors; The multiple clothing order production scheduling text embedding vectors are passed through a clothing order production scheduling text feature encoder based on multiple convolutional layers to obtain the clothing order production scheduling text feature vector.

5. The dynamic scheduling management method for garment production according to claim 4, characterized in that: Associating the clothing order production process associated feature vector with the clothing order production scheduling text feature vector to obtain the clothing order multimodal information associated feature vector, including: Install the Java development environment JDK and necessary dependent libraries for vector computing; Create a ServerSocket instance named MultiModalFeatureServer, listen to the specified port, and wait for client connections; When the client connects to the server, ServerSocket returns a Socket object through which communication with the client is carried out; Receive and parse the garment order production process association feature vector and the garment order production scheduling text feature vector from the client; Use Apache Commons Math to calculate the product of the two feature vectors, and send the calculated clothing order multimodal information association feature vector back to the client; After the calculation is completed, close the Socket connection and release resources.

6. The dynamic scheduling management method for garment production according to claim 5, characterized in that: Extracting features of the clothing production equipment location information to obtain a semantic feature vector of the clothing production equipment location information includes: Segmenting the clothing production equipment location information to obtain a clothing production equipment location information word sequence; The clothing production equipment location information word sequence is passed through a clothing production equipment location information semantic encoder based on a context encoder to obtain the clothing production equipment location information semantic feature vector.

7. The dynamic scheduling management method for garment production according to claim 6, characterized in that: Based on the garment order multimodal information association feature vector and the garment production equipment location information semantic feature vector, a garment order production process scheduling table is generated, including: Fusion of the garment order multimodal information association feature vector and the garment production equipment location information semantic feature vector to obtain a garment order scheduling generation feature vector; Performing feature modulation optimization based on learning parameter structure redundancy on the clothing order scheduling generation feature vector to obtain a sparse optimized clothing order scheduling generation feature vector; The sparse optimization clothing order scheduling generates a feature vector which is passed through a generator to generate a clothing order production process scheduling table.

8. The dynamic scheduling management method for garment production according to claim 7, characterized in that: The clothing order scheduling generation feature vector is optimized by feature modulation based on learning parameter structure redundancy to obtain a sparse optimized clothing order scheduling generation feature vector, including: Extracting each row vector of the learning parameter matrix in the generator as a set of row vectors of clothing order scheduling learning parameters; Calculating the clothing order scheduling entropy degradation maximization simulation value between the clothing order scheduling generation feature vector and each clothing order scheduling learning parameter row vector in the set of clothing order scheduling learning parameter row vectors to obtain a set of clothing order scheduling entropy degradation maximization simulation values; Based on the comparison between each clothing order scheduling entropy degradation maximization simulation value in the set of clothing order scheduling entropy degradation maximization simulation values ​​and a preset threshold, the learning parameter matrix is ​​gated and pruned to obtain a clothing order scheduling learning parameter entropy degradation optimization matrix; The clothing order scheduling generated feature vector is matrix multiplied with the clothing order scheduling learning parameter entropy degradation optimization matrix to obtain the sparse optimized clothing order scheduling generated feature vector.

9. A dynamic scheduling management system for garment production, characterized in that: include: The clothing production information acquisition module is used to obtain the production process data of clothing orders, the production scheduling data of clothing orders and the location information of clothing production equipment; A clothing production information extraction module, used to extract a clothing order multimodal information association feature vector and a clothing production equipment location information semantic feature vector from the production process data of the clothing order, the production scheduling data of the clothing order and the clothing production equipment location information; A clothing order production process scheduling table generation module is used to generate a clothing order production process scheduling table based on the clothing order multimodal information association feature vector and the clothing production equipment location information semantic feature vector.

10. The dynamic scheduling management system for garment production according to claim 9, characterized in that: Garment production information extraction module, including: Extracting a garment order multimodal information association feature vector and a garment production equipment location information semantic feature vector from the garment order production process data, the garment order production schedule data and the garment production equipment location information, including: Extracting features from the production process data of the clothing order to obtain a clothing order production process-related feature vector; Performing feature extraction on the production scheduling data of the clothing order to obtain a clothing order production scheduling text feature vector; Associating the clothing order production process associated feature vector with the clothing order production scheduling text feature vector to obtain the clothing order multimodal information associated feature vector; Feature extraction is performed on the clothing production equipment location information to obtain a semantic feature vector of the clothing production equipment location information.

Images