A clothing production capacity prediction system and method

Abstract

The present invention relates to the technical field of capacity prediction, and proposes a clothing capacity prediction system and method, which collects clothing production data of several process areas; converts raw material pre-processing data, assembly line processing data and manual re-inspection data into trend combination time series features, and predicts clothing capacity estimation value together with raw material preparation data and delivery time; finally, according to the clothing capacity estimation value and clothing capacity preset value, judges whether the clothing capacity meets the requirements, and generates a capacity optimization strategy. Thus, the production data of different production links are integrated, the periodic law of the processing process is obtained from the frequency perspective, the correlation between different links is mined through the constructed neural network, and finally the capacity estimation value is obtained by combining the production constraints through the multi-layer perceptron to guide production optimization, which solves the technical problem that the existing capacity estimation method based on linear regression cannot have a good effect on clothing production.

Description

Technical Field

[0001] The present invention relates to the technical field of production capacity forecasting, and in particular to a clothing production capacity forecasting system and method. Background Art

[0002] With the rapid development of the manufacturing industry, real-time capacity estimation management systems have become particularly important for production managers, as they can help companies achieve refined management and optimization of the production process and improve production efficiency. However, capacity is affected by many factors, and general production management systems often rely on manual statistics of various factors and input them into the system, which cannot accurately estimate capacity and results in large errors. Moreover, such production systems cannot monitor and estimate in real time and have large delays. Therefore, in order to improve production efficiency, reasonably allocate resources, and avoid product delivery delays, it is crucial to study a low-latency capacity estimation method and management system.

[0003] Most of the existing capacity estimation methods are based on linear regression. Linear regression analysis is usually based on the assumption of linear relationship, that is, it is assumed that there is a linear relationship between the independent variable and the dependent variable. This type of method is suitable for scenarios with relatively fixed production patterns, such as vehicle production and electronic product production. Usually, the production lines of these products are independent, that is, one production line only produces one product. Even if some products of the same category may be produced on the same production line, the similarity of these products is extremely high, such as different generations of the same mobile phone. The fixed attributes of the product and the independence of the production line make the data type and its characteristics relatively stable, so the use of linear regression-based methods can achieve better results. However, there are more uncertainties in the clothing production process. First, the production process of clothing is universal in the production of different categories of clothing, that is, one production line may be responsible for the production of multiple types of clothing, such as underwear, socks, coats, etc.; second, clothing production is more prone to rework problems, such as irregular seams. These garments are usually immediately sent back to the corresponding processing link for reprocessing after being found to be abnormal. Therefore, the existing linear regression-based methods cannot achieve good results in the estimation of clothing production capacity. Summary of the invention

[0004] The present invention provides a clothing production capacity prediction system and method, aiming to solve the above-mentioned prior art problems.

[0005] A first aspect of the present invention provides a clothing production capacity prediction system, comprising:

[0006] The edge computing module is arranged in several process areas of the clothing production line in the form of distributed nodes and is configured to collect clothing production data of the several process areas;

[0007] The clothing production data includes raw material preparation data, raw material pre-processing data, assembly line processing data and manual re-inspection data;

[0008] A capacity estimation module receives the garment production data transmitted by the edge computing module, and is configured to convert raw material pre-processing data, assembly line processing data, and manual re-inspection data into trend combination time series features, and predict the estimated value of garment capacity based on the trend combination time series features, the raw material preparation data, and the delivery time;

[0009] The production optimization module receives the clothing production capacity estimation value transmitted by the clothing production capacity estimation module and is configured to determine whether the clothing production capacity meets the requirements based on the clothing production capacity estimation value and the clothing production capacity preset value. If not, a production capacity optimization strategy is generated based on the clothing production data.

[0010] Optionally, the edge computing module specifically includes:

[0011] A raw material preparation data acquisition unit is configured to call a raw material input / output database to obtain raw material input / output data stored in the raw material input / output database; wherein the raw material input / output data includes an estimated raw material usage, a current raw material inventory, and an estimated raw material purchase quantity for each order;

[0012] The raw material preparation data preprocessing unit is configured to classify the acquired raw material input and output data according to each order and generate structured data indexed by ID.

[0013] Optionally, the edge computing module further includes:

[0014] The raw material pre-processing data collection unit is configured to collect the actual processing time of each batch of raw materials at different stages of the garment processing process;

[0015] The raw material pre-processing data pre-processing unit is configured to calculate the processable quantity of each raw material in each batch of orders per unit time according to the actual processing time and the estimated raw material usage.

[0016] Optionally, the edge computing module further includes:

[0017] The assembly line processing data acquisition unit is configured to collect processing operation video data of each processing position on the assembly line;

[0018] The assembly line processing data preprocessing unit is configured to perform standardized identification of worker operations in the collected processing operation video data, record the number of unqualified operations and the number of processes performed by each workstation in the current batch processing unit time.

[0019] Optionally, the edge computing module further includes:

[0020] Manual re-inspection data collection unit;

[0021] The manual re-inspection data collection unit is configured to collect the number of unqualified items and unqualified links obtained through manual re-inspection.

[0022] Optionally, the production capacity estimation module specifically includes:

[0023] The feature extraction unit is configured to encode the time series data composed of raw material pre-processing data, assembly line processing data and manual re-inspection data in time sequence by using a BERT encoder, and convert the encoded data into trend combination time series features;

[0024] The clothing production capacity prediction unit is configured to determine the correlation weights between different production links as output according to the trend combination time series characteristics, and take the real part of the output and splice it with the raw material preparation data and delivery time, and then input it into the multi-layer perceptron to obtain the clothing production capacity estimation value output by the multi-layer perceptron.

[0025] Optionally, the encoded data is converted into trend combination time series features, specifically:

[0026] Perform discrete Fourier transform on the encoded data to obtain trend combination time series characteristics;

[0027] Among them, the expression of the transformed trend combination time series characteristics is specifically:

[0028] ;

[0029] in, is the frequency, is the encoded data, is the integral variable, j is the imaginary unit, T is the upper limit of the integral time interval, Indicates the integral variable representation of the encoded data during discrete Fourier transform, , , They respectively represent the trend combination time series characteristics corresponding to the transformation of raw material pre-processing data, assembly line processing data and manual re-inspection data.

[0030] Optionally, according to the trend combination time series characteristics, the association weights between different production links are determined as output, specifically:

[0031] Will , , Split into multiple patches according to the preset scale, each patch contains data at multiple time points;

[0032] Each Patch is divided into , , The combined data are sent to a fitting network consisting of 2 linear layers and 1 convolutional layer to generate the association weights between different production links.

[0033] Optionally, based on the clothing production data, a production capacity optimization strategy is generated, specifically:

[0034] Determine whether the estimated raw material usage is greater than the sum of the current raw material inventory and the estimated raw material purchase volume; if so, generate raw material preparation optimization suggestions based on the raw material difference; if not, obtain manual re-inspection data;

[0035] Determine whether there are unqualified links in the manual re-inspection data with the number of unqualified links exceeding the preset number; if so, generate assembly line processing optimization suggestions based on the production data of each workstation in the unqualified link; if not, generate raw material pre-processing optimization suggestions or assembly line scheduling optimization suggestions based on the raw material pre-processing data.

[0036] A second aspect of the present invention provides a method for predicting clothing production capacity, comprising:

[0037] Collect clothing production data from several process areas;

[0038] The clothing production data includes raw material preparation data, raw material pre-processing data, assembly line processing data and manual re-inspection data;

[0039] Convert raw material pre-processing data, assembly line processing data and manual re-inspection data into trend combination time series features, and predict the estimated value of clothing production capacity based on the trend combination time series features, the raw material preparation data and delivery time;

[0040] According to the estimated value of clothing production capacity and the preset value of clothing production capacity, it is judged whether the clothing production capacity meets the requirements. If not, a production capacity optimization strategy is generated based on the clothing production data.

[0041] The beneficial effects of the present invention are as follows: a clothing production capacity prediction system and method are proposed, which collects clothing production data of several process areas including raw material preparation data, raw material pre-processing data, assembly line processing data and manual re-inspection data; converts clothing production data into trend combination time series features, and predicts clothing production capacity estimation based on the trend combination time series features, the raw material preparation data and delivery time; finally, according to the clothing production capacity estimation value and the clothing production capacity preset value, it is judged whether the clothing production capacity meets the requirements, if not, based on the clothing production data, a production capacity optimization strategy is generated. Thus, the production data of different production links are fused, the periodic law of the processing process is obtained from the frequency perspective, the correlation between different links is mined through the constructed neural network, and finally the production capacity estimation value is obtained by combining the production constraints through the multi-layer perceptron to guide production optimization, which solves the technical problem that the existing linear regression-based production capacity estimation method cannot have a good effect on clothing production. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 Schematic diagram of the structure of the clothing production capacity prediction system of this embodiment.

[0043] Figure 2 This is a schematic diagram of combining each patch in sequence in parallel in this embodiment.

[0044] Figure 3 Schematic diagram of the flow of the clothing production capacity prediction method of this embodiment.

[0045] Reference numerals:

[0046] 10-Edge computing module; 20-Capacity estimation module; 30-Production optimization module. DETAILED DESCRIPTION

[0047] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0048] Embodiment 1:

[0049] like Figure 1As shown, a clothing production capacity prediction system includes: an edge computing module 10, which is set in several process areas of a clothing production line in the form of a distributed node, and is configured to collect clothing production data from several process areas; wherein the clothing production data includes raw material preparation data, raw material pre-processing data, assembly line processing data and manual re-inspection data; a capacity estimation module 20, which receives the clothing production data transmitted by the edge computing module 10, and is configured to convert the raw material pre-processing data, assembly line processing data and manual re-inspection data into trend combination time series features, and predict the clothing production capacity estimation value based on the trend combination time series features, the raw material preparation data and the delivery time; a production optimization module 30, which receives the clothing production capacity estimation value transmitted by the capacity estimation module 20, and is configured to judge whether the clothing production capacity meets the requirements according to the clothing production capacity estimation value and the clothing production capacity preset value, and if not, generate a capacity optimization strategy based on the clothing production data.

[0050] It should be noted that most of the existing capacity estimation methods are based on linear regression. Linear regression analysis is usually based on the assumption of a linear relationship, that is, it is assumed that there is a linear relationship between the independent variable and the dependent variable. However, there are many uncertainties in the clothing production process. First, the production process of clothing is universal for the production of different categories of clothing, that is, a production line may be responsible for the production of a variety of clothing, such as underwear, socks, jackets, etc.; second, clothing production is more prone to rework problems, such as irregular seams. These garments are usually immediately sent back to the corresponding processing link for reprocessing after being found to be abnormal. Therefore, the existing methods based on linear regression cannot have a good effect on the capacity estimation of clothing production.

[0051] In order to solve the above problems, this embodiment integrates the production data of different production links, obtains the periodic law of the processing process from the perspective of frequency, mines the correlation between different links through the constructed neural network, and finally obtains the capacity estimation value through the multi-layer perceptron combined with production constraints to guide production optimization, which solves the technical problem that the existing capacity estimation method based on linear regression cannot achieve good results in clothing production.

[0052] In a preferred embodiment, the edge computing module 10 specifically includes: a raw material preparation data acquisition unit, configured to call the raw material input and output database to obtain the raw material input and output data stored in the raw material input and output database; wherein the raw material input and output data includes the estimated raw material usage, current raw material inventory and estimated raw material purchase quantity of each order; a raw material preparation data preprocessing unit, configured to classify the acquired raw material input and output data according to each order, and generate structured data indexed by ID.

[0053] On this basis, the edge computing module 10 also includes: a raw material pre-processing data collection unit, which is configured to collect the actual processing time of each batch of raw materials in different links of various processes in the clothing processing process; a raw material pre-processing data preprocessing unit, which is configured to calculate the processable quantity of each raw material in each batch of orders per unit time based on the actual processing time and the estimated raw material usage.

[0054] On this basis, the edge computing module 10 also includes: an assembly line processing data acquisition unit, which is configured to collect processing operation video data of each processing position on the assembly line; an assembly line processing data preprocessing unit, which is configured to perform standardized identification of worker operations in the collected processing operation video data, and record the number of unqualified operations and the number of processes performed by each workstation in the current batch processing unit time.

[0055] On this basis, the edge computing module 10 further includes: a manual re-inspection data collection unit; wherein the manual re-inspection data collection unit is configured to collect the number of unqualified items and unqualified links obtained by manual re-inspection.

[0056] In this embodiment, the edge computing module 10 is mainly deployed in the form of a distributed node, and collects clothing production data from several process areas; among them, the clothing production data includes data from the following four sections in the clothing production process: raw material preparation, raw material pre-processing, assembly line processing and manual re-inspection.

[0057] (1) For raw material preparation data, the edge computing module 10 deployed by the raw material preparation section is responsible for collecting raw material in-and-out data. Usually, there are detailed records of raw material out-and-in. The in-and-out database includes the estimated raw material usage, current raw material inventory, and estimated raw material purchase volume (by category) for each order. After the collection is completed, the collected data is classified according to each order, as shown in the following table, thereby generating a series of structured data indexed by ID.

[0058] ID (Batch) Estimated usage Current inventory Estimated purchase volume category 00000001 20000 4000000000 30000 cotton

[0059] (2) For raw material pre-processing, raw material pre-processing includes multiple processes, such as shrinkage treatment, printing and dyeing finishing, etc. The edge computing module 10 deployed in the raw material pre-processing section is responsible for collecting the actual processing time of each process for each batch of raw materials in different links, and then calculates the processable quantity of each raw material in each batch of orders per unit time based on the collected time and the material usage collected in the raw material preparation stage.

[0060] (3) For assembly line processing, the assembly line processing links need to be distinguished. There are differences in the processing links of different categories of clothing. For example, an assembly line has cutting, sewing, and ironing functions. When processing underwear, cutting and sewing functions will be used, but underwear does not need ironing. If it is used to process shirts, all three processes are required. The implementation of clothing assembly line processes still mostly needs to be completed manually. Therefore, whether the operations of analysts during process operations meet the standards is related to the product qualification rate in the subsequent manual re-inspection stage. The edge computing module 10 deployed in the assembly line processing link uses a camera to continuously acquire video data of the processing operations of each processing position, and then identifies whether the workers' operations in the video are standardized (recording the number of unqualified operations) and simultaneously records the number of processes (processing speed) of each workstation under the current batch processing unit time. The recorded content is shown in the following table.

[0061] ID (Batch) Tailoring Cutting speed Sewing Sewing speed … 00000001 1 20 2 10 …

[0062] (4) For manual re-inspection, the edge computing deployment of the manual re-inspection link is relatively simple. It only needs to record the number of unqualified products and unqualified links, and feed the data back to the production optimization module.

[0063] Therefore, this embodiment can help extract the correlation between different links in the capacity prediction step based on the current production behavior by acquiring the production data of different production links, and has better prediction accuracy than the traditional linear regression method.

[0064] In a preferred embodiment, the capacity estimation module 20 specifically includes: a feature extraction unit, configured to use a BERT encoder to encode time series data consisting of raw material pre-processing data, assembly line processing data and manual re-inspection data in chronological order, and convert the encoded data into trend combination time series features; a clothing capacity prediction unit, configured to determine the correlation weights between different production links as output based on the trend combination time series features, and take the real part of the output and splice it with the raw material preparation data and delivery time, and then input it into a multi-layer perceptron to obtain the clothing capacity estimation value output by the multi-layer perceptron.

[0065] Furthermore, the encoded data is converted into trend combination time series features, specifically: the encoded data is subjected to discrete Fourier transform to obtain trend combination time series features; wherein the expression of the transformed trend combination time series features is specifically:

[0066] ;

[0067] in, is the frequency, is the encoded data, is the integral variable, j is the imaginary unit, T is the upper limit of the integral time interval, Indicates the integral variable representation of the encoded data during discrete Fourier transform, , , They respectively represent the trend combination time series characteristics corresponding to the transformation of raw material pre-processing data, assembly line processing data and manual re-inspection data.

[0068] On this basis, according to the trend combination time series characteristics, the association weights between different production links are determined as output, specifically: , , Split into multiple patches according to the preset scale, each patch contains data at multiple time points; divide each patch according to , , The combined data are sent to a fitting network consisting of 2 linear layers and 1 convolutional layer to generate the association weights between different production links.

[0069] In this embodiment, the feature extraction unit first summarizes the output results of the edge computing module 10, including raw material preparation data, raw material pre-processing data, assembly line processing data, and manual re-inspection data. The raw material preparation data is used as a constraint because the raw materials consumed in the production of clothing should be equal to or less than the raw material reserves, which is the current storage plus the raw materials expected to be purchased within the production time. In addition to the raw material preparation data, other data are sequenced in order of collection time to form time series data (three groups: A, B, C), and then the feature extraction unit encodes each item in the time series data. The encoding can be completed using some pre-trained encoders, such as BERT. After the encoding is completed, the time series data is discrete Fourier transformed, which is conducive to identifying periodic or trend patterns in the data.

[0070] After that, the raw material pre-processing data, assembly line processing data and manual re-inspection data are transformed into the corresponding trend combination time series features , , It is split into multiple patches according to a certain scale. A patch contains data at multiple time points. , , The order is combined side by side, such as Figure 2As shown; finally, the combined data is sent to a fitting network consisting of 2 linear layers and 1 convolutional layer to generate the association weights between different production links; the association weights between different production links are taken as the real part of the output and then concatenated with the raw material preparation data and the delivery time (how long it takes to complete the production) (for the purpose of serving as a constraint) and sent to the multi-layer perceptron to finally obtain the estimated value of the clothing production capacity.

[0071] Therefore, in order to overcome the drawbacks of the existing linear regression-based method in clothing production, this embodiment solves the difficulty of producing different categories of clothing on the same line by dividing the data of different links into multiple patches and fitting the association weights between different patches using a neural network. In addition, the motivation for using Fourier transform to obtain periodic information is that the different data performances in each link of the processing process should be periodic. For example, underwear processing will not generate data in ironing, but shirts will. This periodic property can help the model better separate the characteristics of different categories, thereby improving the estimation accuracy.

[0072] In a preferred embodiment, a production capacity optimization strategy is generated based on the clothing production data, specifically: determine whether the expected raw material usage is greater than the sum of the current raw material inventory and the expected raw material purchase amount; if so, generate raw material preparation optimization suggestions based on the raw material difference; if not, obtain manual re-inspection data; determine whether there is an unqualified link in the manual re-inspection data with the number of unqualified links exceeding a preset number; if so, generate assembly line processing optimization suggestions based on the production data of each workstation in the unqualified link; if not, generate raw material pre-processing optimization suggestions or assembly line scheduling optimization suggestions based on the raw material pre-processing data.

[0073] In this embodiment, the production optimization module 30 first obtains the analysis result of the capacity estimation module 20, which is obtained based on the current production behavior that has occurred, and reflects the possibility of whether the current production process can meet the target capacity. When the capacity estimation is lower than the target value, it is first determined whether the raw materials are sufficient. If they are sufficient, the manual re-inspection results are obtained to determine whether there are serious unqualified conditions. If so, the information is fed back to the unqualified link, and the production data of each station in the link is retrieved to correct the problematic station. If the manual re-inspection is normal, the possibility that the pre-processing output of the raw materials is lower than the production line output is considered, and the pre-processing efficiency is appropriately improved. If the demand is still not met after the increase, a new assembly line is appropriately added for the batch production. Thus, the production data of different production links are fused, the periodic law of the processing process is obtained from the perspective of frequency, the correlation between different links is mined through the constructed neural network, and finally the capacity estimation value is obtained by combining the multi-layer perceptron with the production constraints to guide production optimization, and provide optimization suggestions for improving clothing production capacity for the current clothing production.

[0074] Embodiment 2:

[0075] Reference Figure 3 , Figure 3 A schematic flow chart of a method for predicting clothing production capacity provided by an embodiment of the present invention.

[0076] like Figure 3 As shown, a clothing production capacity prediction method is used in the clothing production capacity prediction system as described above, comprising:

[0077] S1: Collect clothing production data from several process areas;

[0078] The clothing production data includes raw material preparation data, raw material pre-processing data, assembly line processing data and manual re-inspection data;

[0079] S2: converting raw material pre-processing data, assembly line processing data and manual re-inspection data into trend combination time series features, and predicting the estimated value of clothing production capacity based on the trend combination time series features, the raw material preparation data and delivery time;

[0080] S3: judging whether the clothing production capacity meets the requirements according to the clothing production capacity estimation value and the clothing production capacity preset value, and if not, generating a production capacity optimization strategy based on the clothing production data.

[0081] This embodiment provides a method for predicting clothing production capacity. The production data of different production links are integrated to obtain the periodic law of the processing process from the perspective of frequency. The correlation between different links is mined through the constructed neural network. Finally, a multi-layer perceptron is used in combination with production constraints to obtain a production capacity estimation value to guide production optimization. This solves the technical problem that the existing linear regression-based production capacity estimation method cannot achieve good results in clothing production.

[0082] The specific implementation of the clothing production capacity prediction method of the present application is basically the same as the above-mentioned clothing production capacity prediction system embodiment, and will not be repeated here.

[0083] In the description of the embodiments of the present invention, it should be understood that the terms "upper", "lower", "front", "back", "left", "right", "vertical", "horizontal", "center", "top", "bottom", "top", "bottom", "inside", "outside", "inner side", "outer side" and the like indicate orientations or positional relationships based on the orientations or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on the present invention. Among them, "inside" refers to an internal or enclosed area or space. "Periphery" refers to the area surrounding a specific component or a specific area.

[0084] In the description of the embodiments of the present invention, the terms "first", "second", "third", and "fourth" are used for descriptive purposes only and are not to be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, a feature defined as "first", "second", "third", and "fourth" may explicitly or implicitly include one or more of the features. In the description of the present invention, unless otherwise specified, "multiple" means two or more.

[0085] In the description of the embodiments of the present invention, it should be noted that, unless otherwise clearly specified and limited, the terms "install", "connect", "connect", and "assemble" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a direct connection, or an indirect connection through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0086] In the description of the embodiments of the present invention, specific features, structures, materials or characteristics may be combined in a suitable manner in any one or more embodiments or examples.

[0087] In the description of the embodiments of the present invention, it should be understood that "-" and "~" represent a range between two values, and the range includes the endpoints. For example: "AB" represents a range greater than or equal to A and less than or equal to B. "A~B" represents a range greater than or equal to A and less than or equal to B.

[0088] In the description of the embodiments of the present invention, the term "and / or" herein is merely a description of the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B may represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " herein generally indicates that the associated objects before and after are in an "or" relationship.

[0089] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A clothing production capacity prediction system, characterized in that: include: The edge computing module is arranged in several process areas of the clothing production line in the form of distributed nodes and is configured to collect clothing production data of the several process areas; The clothing production data includes raw material preparation data, raw material pre-processing data, assembly line processing data and manual re-inspection data; A capacity estimation module receives the garment production data transmitted by the edge computing module, and is configured to convert raw material pre-processing data, assembly line processing data, and manual re-inspection data into trend combination time series features, and predict the estimated value of garment capacity based on the trend combination time series features, the raw material preparation data, and the delivery time; The production capacity estimation module specifically includes: The feature extraction unit is configured to encode the time series data composed of raw material pre-processing data, assembly line processing data and manual re-inspection data in time sequence by using a BERT encoder, and convert the encoded data into trend combination time series features; A clothing production capacity prediction unit is configured to determine the association weights between different production links as output according to the trend combination time series characteristics, and take the real part of the output, splice it with the raw material preparation data and the delivery time, and then input it into a multi-layer perceptron to obtain the clothing production capacity estimation value output by the multi-layer perceptron; Among them, the encoded data is converted into trend combination time series features, specifically: Perform discrete Fourier transform on the encoded data to obtain trend combination time series characteristics; Among them, the expression of the transformed trend combination time series characteristics is specifically: ； in, is the frequency, is the encoded data, is the integral variable, j is the imaginary unit, T is the upper limit of the integral time interval, Indicates the integral variable representation of the encoded data during discrete Fourier transform, , , They respectively represent the trend combination time series characteristics corresponding to the transformation of raw material pre-processing data, assembly line processing data and manual re-inspection data; According to the trend combination time series characteristics, the association weights between different production links are determined as output, specifically: Will , , Split into multiple patches according to the preset scale, each patch contains data at multiple time points; Each Patch is divided into , , The combined data are sent to a fitting network consisting of 2 linear layers and 1 convolutional layer to generate the association weights between different production links; The production optimization module receives the clothing production capacity estimation value transmitted by the clothing production capacity estimation module and is configured to determine whether the clothing production capacity meets the requirements based on the clothing production capacity estimation value and the clothing production capacity preset value. If not, a production capacity optimization strategy is generated based on the clothing production data.

2. The clothing production capacity prediction system according to claim 1, characterized in that: The edge computing module specifically includes: A raw material preparation data acquisition unit is configured to call a raw material input / output database to obtain raw material input / output data stored in the raw material input / output database; wherein the raw material input / output data includes an estimated raw material usage, a current raw material inventory, and an estimated raw material purchase quantity for each order; The raw material preparation data preprocessing unit is configured to classify the acquired raw material input and output data according to each order and generate structured data indexed by ID.

3. The clothing production capacity prediction system according to claim 2, characterized in that: The edge computing module further includes: The raw material pre-processing data collection unit is configured to collect the actual processing time of each batch of raw materials at different stages of the garment processing process; The raw material pre-processing data pre-processing unit is configured to calculate the processable quantity of each raw material in each batch of orders per unit time according to the actual processing time and the estimated raw material usage.

4. The clothing production capacity prediction system according to claim 3, characterized in that: The edge computing module further includes: The assembly line processing data acquisition unit is configured to collect processing operation video data of each processing position on the assembly line; The assembly line processing data preprocessing unit is configured to perform standardized identification of worker operations in the collected processing operation video data, record the number of unqualified operations and the number of processes performed by each workstation in the current batch processing unit time.

5. The clothing production capacity prediction system according to claim 4, characterized in that: The edge computing module further includes: Manual re-inspection data collection unit; The manual re-inspection data collection unit is configured to collect the number of unqualified items and unqualified links obtained through manual re-inspection.

6. The clothing production capacity prediction system according to claim 5, characterized in that: Based on the clothing production data, a production capacity optimization strategy is generated, specifically: Determine whether the estimated raw material usage is greater than the sum of the current raw material inventory and the estimated raw material purchase volume; if so, generate raw material preparation optimization suggestions based on the raw material difference; if not, obtain manual re-inspection data; Determine whether there are unqualified links in the manual re-inspection data with a number of unqualified links exceeding a preset number; If so, then based on the production data of each workstation in the unqualified link, an optimization suggestion for assembly line processing is generated; if not, based on the raw material pre-processing data, an optimization suggestion for raw material pre-processing or an optimization suggestion for assembly line scheduling is generated.

7. A clothing production capacity prediction method, characterized in that: The clothing production capacity prediction system according to any one of claims 1 to 6 comprises: Collect clothing production data from several process areas; The clothing production data includes raw material preparation data, raw material pre-processing data, assembly line processing data and manual re-inspection data; Convert raw material pre-processing data, assembly line processing data and manual re-inspection data into trend combination time series features, and predict the estimated value of clothing production capacity based on the trend combination time series features, the raw material preparation data and delivery time; According to the estimated value of clothing production capacity and the preset value of clothing production capacity, it is judged whether the clothing production capacity meets the requirements. If not, a production capacity optimization strategy is generated based on the clothing production data.

Images