Sewing production time prediction method, device and equipment and storage medium

Abstract

The invention discloses a sewing production time prediction method and device, equipment and a storage medium, and relates to the technical field of garment manufacturing, and the method comprises the steps: carrying out the data cleaning of historical production data of a garment production system, so as to obtain the cleaned production data; determining each historical efficiency score of each worker for each sewing process based on worker historical operation data in the obtained cleaned production data, and constructing a corresponding matrix by using each historical efficiency score to obtain a worker skill matrix; and determining a training set and a test set based on the cleaned production data, training and testing the initial sewing man-hour prediction model by using the training set and the test set, and determining a sewing man-hour prediction result corresponding to the sewing man-hour prediction request based on an obtained target sewing man-hour prediction model. Therefore, insufficient production scheduling precision caused by incapability of adapting to individual differences of employees during sewing time prediction can be avoided.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of garment manufacturing, and in particular relates to a sewing production working hour prediction method, device, equipment and storage medium. BACKGROUND

[0002] In the garment manufacturing industry, production scheduling usually relies on standard working hours for sewing process allocation, but actual working hours of employees differ due to factors such as skills and proficiency, resulting in a large deviation between scheduling plans and actual execution, frequent adjustment of production lines, replacement of employees or rearrangement of sewing processes, which seriously affects production efficiency. Traditional methods rely on manual experience to estimate working hours, lack data support, and are difficult to accurately match employee capabilities and sewing process requirements, resulting in waste of production capacity or production delay. In the prior art, sewing production working hour prediction mostly uses fixed standard values or simple statistical methods, which cannot dynamically adapt to individual differences of employees, resulting in limited scheduling optimization effect.

[0003] From the above, how to avoid the problem that the individual differences of employees cannot be adapted during sewing working hour prediction, resulting in insufficient scheduling accuracy is an urgent problem to be solved at present. SUMMARY

[0004] Therefore, the purpose of the present application is to provide a sewing production working hour prediction method, device, equipment and storage medium, which can avoid the problem that the individual differences of employees cannot be adapted during sewing working hour prediction, resulting in insufficient scheduling accuracy. The specific scheme is as follows:

[0005] In a first aspect, the present application provides a sewing production working hour prediction method, comprising:

[0006] performing data cleaning on historical production data of a garment production system to obtain cleaned production data;

[0007] determining historical efficiency scores of each worker for each sewing process based on worker historical operation data in the cleaned production data, normalizing each historical efficiency score, and using the obtained normalized efficiency scores to construct a corresponding matrix to obtain a worker skill matrix;

[0008] performing feature extraction on the cleaned production data to obtain multi-dimensional feature data, determining a training set and a test set based on the multi-dimensional feature data, then training and testing an initial sewing working hour prediction model using the training set and the test set to obtain a target sewing working hour prediction model, and determining a sewing working hour prediction result corresponding to a sewing working hour prediction request based on the target sewing working hour prediction model; the initial sewing working hour prediction model is a sewing working hour prediction model determined based on the worker skill matrix and a light gradient boosting machine.

[0009] Optionally, the historical production data of the garment production system is subjected to data cleaning to obtain cleaned production data, comprising:

[0010] The historical production data sent by the hanging software or the garment manufacturing execution system is received by using an asynchronous message queue;

[0011] The target local data of each worker for each sewing process per day is determined based on the historical production data, and all the working hour data of each worker for each sewing process is determined as target global data, and then the target abnormal value range is determined by using the interquartile range filtering method and based on the target local data and the target global data, and the historical production data exceeding the target abnormal value range is removed to obtain cleaned production data.

[0012] Optionally, the target local data of each worker for each sewing process per day is determined based on the historical production data, and all the working hour data of each worker for each sewing process is determined as target global data, and then the target abnormal value range is determined by using the interquartile range filtering method and based on the target local data and the target global data, and the historical production data exceeding the target abnormal value range is removed to obtain cleaned production data, comprising:

[0013] The target local data of each worker for each sewing process per day is determined based on the historical production data, and the first quartile and the third quartile corresponding to the target local data are determined;

[0014] The first interquartile range is determined by using the first quartile and the third quartile corresponding to the target local data, and the first abnormal value range is determined based on the first interquartile range;

[0015] All the working hour data of each worker for each sewing process is determined based on the historical production data, and the target global data is determined by using the all the working hour data, and the first quartile and the third quartile corresponding to the target global data are determined;

[0016] The second interquartile range is determined by using the first quartile and the third quartile corresponding to the target global data, and the second abnormal value range is determined based on the second interquartile range;

[0017] The historical production data exceeding the first abnormal value range and the second abnormal value range is removed to obtain cleaned production data.

[0018] Optionally, the historical efficiency scores of each worker for each sewing process are determined based on the worker historical operation data in the cleaned production data, and the historical efficiency scores are normalized to construct a corresponding matrix by using the obtained normalized efficiency scores to obtain a worker skill matrix, comprising:

[0019] determine a worker set, a sewing procedure set, a sewing procedure standard man-hour set and a sewing procedure production piece number set based on the worker historical operation data in the cleaned production data;

[0020] determine each historical efficiency score of each worker for each sewing procedure by using the worker set, the sewing procedure set and the sewing procedure production piece number set;

[0021] perform standard deviation normalization on each historical efficiency score to obtain a normalized efficiency score, and construct a corresponding matrix by using the normalized efficiency score to obtain a worker skill matrix.

[0022] Optionally, the step of determining each historical efficiency score of each worker for each sewing procedure by using the worker set, the sewing procedure set and the sewing procedure production piece number set comprises:

[0023] determine an actual production time consumption of each worker for each sewing procedure by using the worker set, the sewing procedure set and the sewing procedure production piece number set;

[0024] determine an actual total time consumption of each worker for each sewing procedure based on the actual production time consumption, and determine a theoretical total time consumption of each worker for each sewing procedure by using the sewing procedure standard man-hour set;

[0025] determine each historical efficiency score of each worker for each sewing procedure based on the actual total time consumption and the theoretical total time consumption.

[0026] Optionally, the step of performing feature extraction on the cleaned production data to obtain multi-dimensional feature data, and determining a training set and a test set based on the multi-dimensional feature data comprises:

[0027] extract a worker feature from the cleaned production data and the worker skill matrix; the worker feature comprises the historical efficiency score, a skill level score and a historical operation piece number;

[0028] extract a sewing procedure feature from the cleaned production data; the sewing procedure feature comprises a production date, an order category, a sewing procedure standard man-hour and a sewing procedure text description; the sewing procedure text description is a text description determined based on a natural language processing technology;

[0029] determine an interaction feature based on the worker feature and the sewing procedure feature, and determine multi-dimensional feature data by using the worker feature, the sewing procedure feature and the interaction feature;

[0030] divide the multi-dimensional feature data based on a preset division ratio to obtain the training set and the test set.

[0031] Optionally, the training and testing of the initial sewing working hour prediction model based on the training set and the test set to obtain a target sewing working hour prediction model, and determining a sewing working hour prediction result corresponding to a sewing working hour prediction request based on the target sewing working hour prediction model, comprises:

[0032] determining an initial sewing working hour prediction model based on the worker skill matrix and the light gradient boosting machine, training the initial sewing working hour prediction model based on the training set, and setting a target hyperparameter in the training process to obtain a trained sewing working hour prediction model;

[0033] determining the mean absolute percentage error corresponding to the trained sewing working hour prediction model based on the test set, and adjusting the trained sewing working hour prediction model based on the mean absolute percentage error to obtain a target sewing working hour prediction model;

[0034] obtaining prediction request data corresponding to a sewing working hour prediction request, and determining a corresponding sewing working hour prediction result based on the prediction request data and using the target sewing working hour prediction model.

[0035] In a second aspect, the present application provides a sewing production working hour prediction device, comprising:

[0036] a data cleaning module for cleaning historical production data of a garment production system to obtain cleaned production data;

[0037] a matrix construction module for determining historical efficiency scores of each worker for each sewing process based on worker historical operation data in the cleaned production data, normalizing each historical efficiency score, and constructing a corresponding matrix using the obtained normalized efficiency score to obtain a worker skill matrix;

[0038] a model training module for extracting features from the cleaned production data to obtain multi-dimensional feature data, determining a training set and a test set based on the multi-dimensional feature data, training and testing an initial sewing working hour prediction model based on the training set and the test set to obtain a target sewing working hour prediction model, and determining a sewing working hour prediction result corresponding to a sewing working hour prediction request based on the target sewing working hour prediction model; the initial sewing working hour prediction model is a sewing working hour prediction model determined based on the worker skill matrix and the light gradient boosting machine.

[0039] In a third aspect, the present application provides an electronic device, comprising:

[0040] a memory for saving a computer program;

[0041] A processor is configured to execute the computer program to implement the sewing production working hour prediction method.

[0042] In a fourth aspect, the present application provides a computer readable storage medium for storing a computer program, wherein the computer program, when executed by a processor, implements the sewing production working hour prediction method.

[0043] The historical production data of the garment production system is cleaned to obtain cleaned production data; historical efficiency scores of each worker for each sewing process are determined based on worker historical operation data in the cleaned production data, the historical efficiency scores are normalized, and a worker skill matrix is obtained by constructing a corresponding matrix using the normalized efficiency scores; multi-dimensional feature data is obtained by feature extraction on the cleaned production data, a training set and a test set are determined based on the multi-dimensional feature data, an initial sewing working hour prediction model is trained and tested using the training set and the test set, a target sewing working hour prediction model is obtained, and a sewing working hour prediction result corresponding to a sewing working hour prediction request is determined based on the target sewing working hour prediction model; and the initial sewing working hour prediction model is a sewing working hour prediction model determined based on the worker skill matrix and a light gradient boosting machine.

[0044] As can be seen from the above, the historical production data of the garment production system is cleaned, which can effectively remove abnormal values and noise in the data. Short-term local cleaning can timely process abnormal data appearing in a short period of time, and long-term global cleaning can remove data points that are abnormal in the entire production cycle. Historical efficiency scores of each worker for each sewing process are determined based on the cleaned worker historical operation data, and normalized, and a worker skill matrix is obtained by constructing a corresponding matrix, which can accurately quantify the skill level of the worker in different sewing processes. The multi-dimensional feature data obtained by feature extraction on the cleaned production data is used to train and test an initial sewing working hour prediction model based on the worker skill matrix and the light gradient boosting machine, and finally a target sewing working hour prediction model is obtained. In this way, the production schedule can be reasonably arranged according to the predicted working hours, the human and equipment resources can be reasonably allocated, production surplus or deficiency can be avoided, and the resource utilization rate can be greatly improved. BRIEF DESCRIPTION OF DRAWINGS

[0045] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description are only embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor based on the provided drawings.

[0046] Figure 1A sewing production time prediction method flowchart is disclosed in the present application.

[0047] Figure 2 A sewing process feature schematic diagram is provided in the present application.

[0048] Figure 3 A sewing production time prediction device structure schematic diagram is disclosed in the present application.

[0049] Figure 4 An electronic device structure diagram is disclosed in the present application. DETAILED DESCRIPTION

[0050] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0051] At present, the production scheduling of garment manufacturing usually relies on standard time for sewing process allocation, mainly relying on manual experience to estimate the working hours, lacking data support, and it is difficult to accurately match the employee's ability and the demand of sewing process, resulting in waste of production capacity or production delay. The existing sewing production time prediction method mainly adopts fixed standard value or simple statistical method, which cannot dynamically adapt to individual differences of employees, resulting in limited optimization effect of production scheduling. Therefore, the present application provides a sewing production time prediction method, which can reasonably arrange the production progress according to the predicted working hours, avoid production surplus or deficiency, reasonably allocate human and equipment resources, avoid production surplus or deficiency, and greatly improve the resource utilization rate.

[0052] Referring to Figure 1 The embodiment of the present application discloses a sewing production time prediction method, which comprises:

[0053] Step S11, data cleaning is performed on the historical production data of the garment production system to obtain cleaned production data.

[0054] In this embodiment, the historical production data is obtained from an external system by using an asynchronous message queue; the external system includes a hanger software and a garment manufacturing execution system (MES); after obtaining the historical production data, target local data and target global data corresponding to the historical production data are determined, and the target local data and the target global data are cleaned to obtain cleaned production data. Specifically, the data cleaning of the historical production data of the garment production system to obtain the cleaned production data includes: receiving the historical production data sent by the hanger software or the garment manufacturing execution system by using the asynchronous message queue; determining the target local data of each worker for each sewing process based on the historical production data, and determining all working hour data of each worker for each sewing process as target global data, and then determining a target abnormal value range by using a quartile range filtering method and based on the target local data and the target global data, and removing the historical production data exceeding the target abnormal value range to obtain the cleaned production data.

[0055] It can be understood that the target local data is the data of each worker for each sewing process per day based on the historical production data, and then the first quartile and the third quartile corresponding to the target local data are determined, that is, the target local data is divided into four equal parts, the first quartile corresponds to the value of the node of one fourth of the four equal parts, that is, 25%; the second quartile corresponds to the value of the node of three fourths of the four equal parts, that is, 75%, and the first quartile range is determined by using the first quartile and the third quartile corresponding to the target local data, the first quartile range corresponds to 50% of the four equal parts, and the range exceeding three times the first quartile range is determined as the first abnormal value range. In a specific embodiment, if the target local data located between the first quartile and the third quartile is [3, 4, 5], it is noted that the data is sorted from small to large, then the first quartile range corresponding to the target local data is IQR = 5-3 = 2; the target local data exceeding ± 3 times IQR is regarded as an abnormal value and is removed. The target global data is the same, and the target global data is all working hour data of each worker for each sewing process.

[0056] Specifically, the method for determining the target local data of each worker for each sewing process based on the historical production data, determining all working hour data of each worker for each sewing process as target global data, determining a target abnormal value range based on the target local data and the target global data by using a quartile range filtering method, and eliminating the historical production data exceeding the target abnormal value range to obtain cleaned production data comprises: determining the target local data of each worker for each sewing process based on the historical production data, and determining the first quartile and the third quartile corresponding to the target local data; determining the first quartile range by using the first quartile and the third quartile corresponding to the target local data, and determining the first abnormal value range based on the first quartile range; determining all working hour data of each worker for each sewing process based on the historical production data, and determining the target global data of the all working hour data, determining the first quartile and the third quartile corresponding to the target global data; determining the second quartile range by using the first quartile and the third quartile corresponding to the target global data, and determining the second abnormal value range based on the second quartile range; and eliminating the historical production data exceeding the first abnormal value range and the second abnormal value range to obtain cleaned production data.

[0057] In step S12, each historical efficiency score of each worker for each sewing process is determined based on the worker historical operation data in the cleaned production data, each historical efficiency score is normalized, a corresponding matrix is constructed by using the obtained normalized efficiency score, and a worker skill matrix is obtained.

[0058] In this embodiment, after obtaining the production data after cleaning, a set of workers, a set of sewing processes, a set of standard working hours for sewing processes, and a set of production quantities for sewing processes are determined based on the historical work data of workers in the production data after cleaning. Then, the set of workers, the set of sewing processes, and the set of production quantities for sewing processes are used to determine the historical efficiency scores of each worker for each sewing process, and the deviation standardization of each historical efficiency score is performed to obtain a normalized efficiency score. The normalized efficiency scores are used to construct a corresponding matrix to obtain the worker skill matrix. Specifically, the process of determining the historical efficiency scores of each worker for each sewing process based on the historical work data of workers in the post-cleaning production data, normalizing the historical efficiency scores, and constructing a corresponding matrix using the normalized efficiency scores to obtain the worker skill matrix includes: determining a set of workers, a set of sewing processes, a set of standard working hours for sewing processes, and a set of production quantities for sewing processes based on the historical work data of workers in the post-cleaning production data; determining the historical efficiency scores of each worker for each sewing process using the set of workers, the set of sewing processes, and the set of production quantities for sewing processes; performing deviation standardization on the historical efficiency scores to obtain normalized efficiency scores, and constructing a corresponding matrix using the normalized efficiency scores to obtain the worker skill matrix.

[0059] It is understood that the actual production time of each worker for each sewing process is determined using the worker set, the sewing process set, and the number of pieces produced by each sewing process. Based on the actual production time, the actual total time for each worker to complete each sewing process is determined; for example, worker A's actual total time for sewing buttons is [3, 4, 5] minutes. In one specific embodiment, the worker set is determined based on the worker's historical work data in the post-cleaning production data. Sewing process assembly Standard working hours for sewing processes Set of production quantities for sewing process Then, using the set of workers, the set of sewing processes, and the set of production counts for each sewing process, the historical efficiency score for each worker for each sewing process is determined; the formula corresponding to the historical efficiency score is as follows:

[0060] ;

[0061] in, For workers In the process Historical efficiency scores; For workers In the process The actual total time spent; For workers In the process theoretical total time consumption on the sewing procedure.

[0062] Specifically, the determining of the historical efficiency score of each worker for each sewing procedure by using the worker set, the sewing procedure set and the sewing procedure production piece number set comprises: determining the actual production time consumption of each worker for each sewing procedure by using the worker set, the sewing procedure set and the sewing procedure production piece number set; determining the actual total time consumption of each worker for each sewing procedure based on the actual production time consumption, and determining the theoretical total time consumption of each worker for each sewing procedure by using the sewing procedure standard man-hour set; determining the historical efficiency score of each worker for each sewing procedure based on the actual total time consumption and the theoretical total time consumption.

[0063] Further, after obtaining each historical efficiency score, Min-Max normalization is performed on each historical efficiency score to obtain a normalized efficiency score, and the corresponding formula of the normalization is as follows:

[0064] ;

[0065] wherein, is the normalized efficiency score; is the historical efficiency score of the worker on the sewing procedure ; is the minimum value in each initial efficiency score; is the maximum value in each initial efficiency score. After obtaining the normalized efficiency score, the normalized efficiency score is arranged into a matrix to obtain a worker skill matrix. The worker skill matrix has workers as rows, has sewing procedures as columns, and each element in the matrix is the normalized efficiency score of the worker for the corresponding sewing procedure.

[0066] Step S13, feature extraction is performed on the cleaned production data to obtain multi-dimensional feature data, a training set and a test set are determined based on the multi-dimensional feature data, then an initial sewing man-hour prediction model is trained and tested by using the training set and the test set to obtain a target sewing man-hour prediction model, and a sewing man-hour prediction result corresponding to a sewing man-hour prediction request is determined based on the target sewing man-hour prediction model; the initial sewing man-hour prediction model is a sewing man-hour prediction model determined based on the worker skill matrix and a light gradient boosting machine.

[0067] In this embodiment, the worker features are extracted from the cleaned production data and the worker skill matrix. The worker features include the historical efficiency score, skill level score, and historical workpiece number of the worker in a specific production process. Then, the sewing process features are extracted from the cleaned production data. The sewing process features include the production date, order category, sewing process standard working hours, and sewing process text description. The sewing process text description is a text description determined based on NLP (Natural Language Processing) text analysis technology, Figure 2 A sewing process feature diagram is provided for this embodiment. The worker features and the sewing process features are combined to obtain interaction features, which mainly reflect the interaction between the worker features and the sewing process features, such as historical efficiency statistics of the worker in a specific order category and fabric type process, etc. By comprehensively considering these features, potential information in the data can be mined to provide more rich input features for the model, thereby improving the prediction accuracy of the model. Then, the worker features, the sewing process features, and the interaction features are used to determine multi-dimensional feature data, and the multi-dimensional feature data is divided according to a preset division ratio to obtain a training set and a test set.

[0068] Specifically, the feature extraction from the cleaned production data to obtain multi-dimensional feature data, and the determination of a training set and a test set based on the multi-dimensional feature data, includes: extracting the worker features from the cleaned production data and the worker skill matrix. The worker features include the historical efficiency score, skill level score, and historical workpiece number. Extracting the sewing process features from the cleaned production data. The sewing process features include the production date, order category, sewing process standard working hours, and sewing process text description. The sewing process text description is a text description determined based on natural language processing technology. The interaction features are determined based on the worker features and the sewing process features, and the multi-dimensional feature data is determined using the worker features, the sewing process features, and the interaction features. The multi-dimensional feature data is divided based on a preset division ratio to obtain a training set and a test set. It is worth mentioning that the preset division ratio can be adjusted according to actual conditions, which is not limited here.

[0069] It can be understood that an initial sewing time prediction model is determined based on the worker skill matrix and LightGBM (i.e., light gradient boosting machine), model hyperparameters such as learning rate, number of leaf nodes, etc. are set based on actual conditions, the initial sewing time prediction model is trained using the training set to learn the relationship between time and features, a trained sewing time prediction model is obtained, and then the trained sewing time prediction model is verified using the test set, and the core index is the mean absolute percentage error (MAPE). In a specific embodiment, if the predicted time of the trained sewing time prediction model is , the actual time is , and the corresponding mean absolute percentage error formula is as follows:

[0070] ;

[0071] Further, after obtaining the mean absolute percentage error of the predicted time and the actual time, the smaller the mean absolute percentage error, the more accurate the trained sewing time prediction model is. The mean absolute percentage error is controlled within a preset error range according to actual conditions, and pre-training is updated every day to adapt to changes in production data to obtain a target sewing time prediction model.

[0072] Specifically, the training and testing of the initial sewing time prediction model using the training set and the test set to obtain a target sewing time prediction model, and determining a sewing time prediction result corresponding to a sewing time prediction request based on the target sewing time prediction model, includes: determining an initial sewing time prediction model based on the worker skill matrix and LightGBM, training the initial sewing time prediction model using the training set, and setting target hyperparameters during training to obtain a trained sewing time prediction model; determining the mean absolute percentage error corresponding to the trained sewing time prediction model using the test set, and adjusting the trained sewing time prediction model based on the mean absolute percentage error to obtain a target sewing time prediction model; obtaining prediction request data corresponding to a sewing time prediction request, and determining a corresponding sewing time prediction result based on the prediction request data and using the target sewing time prediction model.

[0073] As can be seen from the above, by receiving historical production data of a garment production system and using quartile range filtering method to perform short-term local cleaning and long-term global cleaning in time dimension, the application can effectively remove abnormal values and noises in the data. The short-term local cleaning can timely process abnormal data appearing in a short time, and the long-term global cleaning can remove data points that are abnormal in the entire production cycle. Based on the cleaned historical work data of workers, the historical efficiency scores of each worker for each sewing process are determined and normalized, and then a corresponding matrix is constructed to obtain a worker skill matrix, which can accurately quantify the skill level of workers in different sewing processes. The cleaned production data is subjected to feature extraction, and the obtained multi-dimensional feature data is used to train and test an initial sewing time prediction model based on the worker skill matrix and the light gradient boosting machine, and finally a target sewing time prediction model is obtained. In this way, the production schedule can be reasonably arranged according to the predicted working hours, the human and equipment resources can be reasonably allocated, the production surplus or deficiency can be avoided, and the resource utilization rate can be greatly improved.

[0074] Correspondingly, referring to Figure 3 The application further provides a sewing production time prediction device, which comprises:

[0075] A data cleaning module 11 is configured to clean the historical production data of the garment production system to obtain cleaned production data.

[0076] A matrix construction module 12 is configured to determine historical efficiency scores of each worker for each sewing process based on historical work data of workers in the cleaned production data, normalize the historical efficiency scores, and construct a corresponding matrix using the normalized efficiency scores to obtain a worker skill matrix.

[0077] A model training module 13 is configured to extract features from the cleaned production data to obtain multi-dimensional feature data, determine a training set and a test set based on the multi-dimensional feature data, train and test an initial sewing time prediction model using the training set and the test set to obtain a target sewing time prediction model, and determine a sewing time prediction result corresponding to a sewing time prediction request based on the target sewing time prediction model. The initial sewing time prediction model is a sewing time prediction model determined based on the worker skill matrix and the light gradient boosting machine.

[0078] From the above, the application can effectively remove abnormal values and noise in the data by receiving historical production data of a garment production system and using quartile range filtering method for short-term local cleaning and long-term global cleaning in time dimension. The short-term local cleaning can timely process abnormal data appearing in a short time, and the long-term global cleaning can remove data points that are abnormal in the entire production cycle. The historical efficiency scores of workers for each sewing process are determined based on the cleaned worker historical operation data, and normalized, and then a corresponding matrix is constructed to obtain a worker skill matrix, which can accurately quantify the skill level of workers in different sewing processes. The cleaned production data is subjected to feature extraction, and the obtained multi-dimensional feature data is used to train and test an initial sewing time prediction model based on the worker skill matrix and the light gradient boosting machine, and finally a target sewing time prediction model is obtained. In this way, the production schedule can be reasonably arranged according to the predicted working hours, the human and equipment resources can be reasonably allocated, the production surplus or deficiency can be avoided, and the resource utilization rate can be greatly improved.

[0079] In some embodiments, the data cleaning module 11 can specifically include:

[0080] The data receiving unit is configured to receive historical production data sent by the hanging software or the garment manufacturing execution system by using an asynchronous message queue.

[0081] The data elimination unit is configured to determine target local data of each worker for each sewing process based on the historical production data, determine all working hour data of each worker for each sewing process as target global data, and then determine a target abnormal value range based on the target local data and the target global data by using the quartile range filtering method, and eliminate the historical production data exceeding the target abnormal value range to obtain cleaned production data.

[0082] In some embodiments, the data cleaning module 11 can specifically include:

[0083] The first quartile determination unit is configured to determine target local data of each worker for each sewing process based on the historical production data, and determine a first quartile and a third quartile corresponding to the target local data.

[0084] The first range determination unit is configured to determine a first quartile range by using the first quartile and the third quartile corresponding to the target local data, and determine a first abnormal value range based on the first quartile range.

[0085] The second quartile determination unit is configured to determine all working hour data of each worker for each sewing process based on the historical production data, and determine the all working hour data as target global data, and determine a first quartile and a third quartile corresponding to the target global data;

[0086] The second range determination unit is configured to determine a second quartile distance by using the first quartile and the third quartile corresponding to the target global data, and determine a second outlier range based on the second quartile distance;

[0087] The production data elimination unit is configured to eliminate the historical production data exceeding the first outlier range and the second outlier range to obtain cleaned production data.

[0088] In some embodiments, the matrix construction module 12 can specifically include:

[0089] The set determination unit is configured to determine a worker set, a sewing process set, a sewing process standard working hour set and a sewing process production piece number set based on worker historical operation data in the cleaned production data;

[0090] The score determination unit is configured to determine each historical efficiency score of each worker for each sewing process by using the worker set, the sewing process set and the sewing process production piece number set;

[0091] The matrix construction unit is configured to perform deviation standardization on each historical efficiency score to obtain a normalized efficiency score, and construct a corresponding matrix by using the normalized efficiency score to obtain a worker skill matrix.

[0092] In some embodiments, the matrix construction module 12 can specifically include:

[0093] The first time consumption determination unit is configured to determine actual production time consumption of each worker for each sewing process by using the worker set, the sewing process set and the sewing process production piece number set;

[0094] The second time consumption determination unit is configured to determine actual total time consumption of each worker for completing each sewing process based on the actual production time consumption, and determine theoretical total time consumption of each worker for completing each sewing process by using the sewing process standard working hour set;

[0095] The efficiency score determination unit is configured to determine each historical efficiency score of each worker for each sewing process based on the actual total time consumption and the theoretical total time consumption.

[0096] In some embodiments, the model training module 13 can specifically include:

[0097] a first feature extraction unit configured to extract worker features from the cleaned production data and the worker skill matrix, the worker features including the historical efficiency score, the skill level score, and the historical piece count;

[0098] a second feature extraction unit configured to extract sewing process features from the cleaned production data, the sewing process features including a production date, an order category, a sewing process standard time, and a sewing process textual description, the sewing process textual description being a textual description determined based on a natural language processing technique;

[0099] a feature data determination unit configured to determine interaction features based on the worker features and the sewing process features, and determine multi-dimensional feature data using the worker features, the sewing process features, and the interaction features;

[0100] a data division unit configured to divide the multi-dimensional feature data based on a preset division ratio to obtain a training set and a test set.

[0101] In some embodiments, the model training module 13 can specifically include:

[0102] a model training unit configured to determine an initial sewing time prediction model based on the worker skill matrix and a light gradient boosting machine, train the initial sewing time prediction model using the training set, and set a target hyperparameter during the training to obtain a trained sewing time prediction model;

[0103] a model adjustment unit configured to determine a mean absolute percentage error corresponding to the trained sewing time prediction model using the test set, and adjust the trained sewing time prediction model based on the mean absolute percentage error to obtain a target sewing time prediction model;

[0104] a prediction result determination unit configured to obtain prediction request data corresponding to a sewing time prediction request, and determine a corresponding sewing time prediction result based on the prediction request data and using the target sewing time prediction model.

[0105] Further, the embodiments of the present application also disclose an electronic device, Figure 4is a structural diagram of the electronic device 20 according to an exemplary embodiment, and the content in the diagram should not be considered as any limitation on the use range of the present application. The electronic device 20 can specifically include at least one processor 21, at least one memory 22, a power supply 23, a communication interface 24, an input / output interface 25, and a communication bus 26. The memory 22 is configured to store a computer program, and the processor 21 is configured to load and execute the computer program to implement the related steps in the sewing production working hour prediction method disclosed in any of the preceding embodiments. In addition, the electronic device 20 in the embodiment can be an electronic computer.

[0106] In the embodiment, the power supply 23 is configured to provide working voltage for each hardware device on the electronic device 20; the communication interface 24 is configured to create a data transmission channel between the electronic device 20 and external devices, and the communication protocol followed by the communication interface 24 can be any communication protocol applicable to the technical solution of the present application, which is not limited here; the input / output interface 25 is configured to obtain external input data or output data to the outside, and the specific interface type can be selected according to the specific application needs, which is not limited here.

[0107] In addition, the memory 22 as a carrier for resource storage can be a read-only memory, a random access memory, a magnetic disk, or an optical disk, etc., and the resources stored thereon can include an operating system 221, a computer program 222, etc., and the storage mode can be temporary storage or permanent storage.

[0108] The operating system 221 is configured to manage and control each hardware device on the electronic device 20 and the computer program 222, and can be Windows Server, Netware, Unix, Linux, etc. The computer program 222 can further include computer programs for completing other specific work in addition to the computer program for completing the sewing production working hour prediction method executed by the electronic device 20 disclosed in any of the preceding embodiments.

[0109] Further, the present application also discloses a computer readable storage medium for storing a computer program; wherein the computer program is executed by a processor to implement the sewing production working hour prediction method disclosed above. For the specific steps of the method, refer to the corresponding content disclosed in the preceding embodiments, which will not be repeated here.

[0110] The embodiments in the specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. For the same or similar parts between the embodiments, refer to each other. For the device disclosed in the embodiments, since it corresponds to the method disclosed in the embodiments, the description is relatively simple, and the relevant parts refer to the method part.

[0111] Those skilled in the art will further appreciate that the units and algorithm steps of the examples described in connection with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or any combination thereof. To clearly illustrate the interchangeability of hardware and software, various components have been described above generally in terms of their functionality, without referring to the details of their implementation. Whether such functionality is implemented in hardware or software depends on the particular application and design constraints imposed on the overall system. Skilled persons can implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present application.

[0112] The steps of a method or algorithm described in connection with the embodiments disclosed herein can be embodied directly in hardware, in a software module executed by a processor, or in a combination of the two. A software module can reside in random access memory (RAM), flash memory, read-only memory (ROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), registers, hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art.

[0113] Finally, it should be noted that the terms "first", "second", and the like, herein do not denote any order, quantity, combination, or importance, but rather are used to distinguish one element from another, and are not necessarily intended to denote the priority of one element over another. Also, the terms "comprises", "comprising", or any other variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but can include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element preceded by "comprises a" does not, without more constraints, foreclose the existence of additional identical elements in the process, method, article, or apparatus that comprises the element.

[0114] The above has introduced the technical solutions provided by the present application in detail, and the principles and implementation manners of the present application have been described by using specific examples. The above description of the examples is only for helping to understand the method of the present application and its core idea; meanwhile, for those skilled in the art, according to the idea of the present application, the specific implementation manner and application range can be changed, and the above description of the present application should not be understood as limiting the present application.

Claims

1. A sewing production time predicting method characterized by comprising: The method comprises the following steps: data cleaning is performed on historical production data of a garment production system to obtain cleaned production data; based on historical work data of workers in the cleaned production data, historical efficiency scores of each worker for each sewing process are determined, and the historical efficiency scores are normalized to obtain normalized efficiency scores, and a worker skill matrix is constructed using the normalized efficiency scores; feature extraction is performed on the cleaned production data to obtain multi-dimensional feature data, a training set and a test set are determined based on the multi-dimensional feature data, and an initial sewing time prediction model is trained and tested using the training set and the test set to obtain a target sewing time prediction model, and a sewing time prediction result corresponding to a sewing time prediction request is determined based on the target sewing time prediction model; the initial sewing time prediction model is a sewing time prediction model determined based on the worker skill matrix and a light gradient boosting machine.

2. The sewing production time predicting method according to claim 1, characterized by, The method comprises the following steps: The historical production data sent by the hanging software or the garment manufacturing execution system is received using an asynchronous message queue; target local data of each worker for each sewing process per day is determined based on the historical production data, all work time data of each worker for each sewing process is determined as target global data, and target outlier ranges are determined using the quartile range filtering method and based on the target local data and the target global data, and the historical production data exceeding the target outlier ranges is removed to obtain cleaned production data.

3. The sewing production time predicting method according to claim 2, characterized by, The method comprises the following steps: target local data of each worker for each sewing process per day is determined based on the historical production data, and first quartiles and third quartiles corresponding to the target local data are determined; a first quartile range is determined using the first quartiles and the third quartiles corresponding to the target local data, and a first outlier range is determined based on the first quartile range; all work time data of each worker for each sewing process is determined based on the historical production data, and the all work time data is determined as target global data, and first quartiles and third quartiles corresponding to the target global data are determined; a second quartile range is determined using the first quartiles and the third quartiles corresponding to the target global data, and a second outlier range is determined based on the second quartile range; the historical production data exceeding the first outlier range and the second outlier range is removed to obtain cleaned production data.

4. The sewing production time predicting method according to claim 1, characterized by, The historical efficiency scores of each worker for each sewing process are determined based on the historical operation data of workers in the cleaned production data, and the historical efficiency scores are normalized to construct a corresponding matrix using the obtained normalized efficiency scores to obtain a worker skill matrix, including: A worker set, a sewing process set, a sewing process standard working hour set, and a sewing process production piece number set are determined based on the historical operation data of workers in the cleaned production data; The historical efficiency scores of each worker for each sewing process are determined using the worker set, the sewing process set, and the sewing process production piece number set; The historical efficiency scores are standardized in deviation to obtain normalized efficiency scores, and a corresponding matrix is constructed using the normalized efficiency scores to obtain a worker skill matrix.

5. The sewing production time predicting method according to claim 4, characterized by, The historical efficiency scores of each worker for each sewing process are determined using the worker set, the sewing process set, and the sewing process production piece number set, including: The actual production time consumption of each worker for each sewing process is determined using the worker set, the sewing process set, and the sewing process production piece number set; The actual total time consumption of each worker for each sewing process is determined based on the actual total time consumption, and the theoretical total time consumption of each worker for each sewing process is determined using the sewing process standard working hour set; The historical efficiency scores of each worker for each sewing process are determined based on the actual total time consumption and the theoretical total time consumption.

6. The sewing production time predicting method according to claim 1, characterized by, The feature extraction is performed on the cleaned production data to obtain multi-dimensional feature data, and a training set and a test set are determined based on the multi-dimensional feature data, including: Worker features are extracted from the cleaned production data and the worker skill matrix; the worker features include the historical efficiency scores, skill level scores, and historical operation piece numbers; Sewing process features are extracted from the cleaned production data; the sewing process features include production dates, order categories, sewing process standard working hours, and sewing process text descriptions; the sewing process text descriptions are text descriptions determined based on natural language processing technology; Interaction features are determined based on the worker features and the sewing process features, and multi-dimensional feature data is determined using the worker features, the sewing process features, and the interaction features; The multi-dimensional feature data is divided based on a preset division ratio to obtain a training set and a test set.

7. The sewing production time predicting method according to any one of claims 1 to 6, characterized in that, The initial sewing working hour prediction model is trained and tested using the training set and the test set to obtain a target sewing working hour prediction model, and a sewing working hour prediction result corresponding to a sewing working hour prediction request is determined based on the target sewing working hour prediction model, including: An initial sewing working hour prediction model is determined based on the worker skill matrix and a light gradient boosting machine, the initial sewing working hour prediction model is trained using the training set, and a target hyperparameter is set during the training process to obtain a trained sewing working hour prediction model; determine a mean absolute percentage error corresponding to the trained sewing time prediction model by using the test set, and adjust the trained sewing time prediction model based on the mean absolute percentage error to obtain a target sewing time prediction model; obtain prediction request data corresponding to a sewing time prediction request, and determine a corresponding sewing time prediction result by using the target sewing time prediction model based on the prediction request data.

8. A sewing production time predicting device characterized by comprising: The method comprises: a data cleaning module configured to clean historical production data of a garment production system to obtain cleaned production data; a matrix construction module configured to determine historical efficiency scores of each worker for each sewing process based on historical work data of the workers in the cleaned production data, normalize each of the historical efficiency scores, and construct a corresponding matrix by using the normalized efficiency scores to obtain a worker skill matrix; a model training module configured to extract features from the cleaned production data to obtain multi-dimensional feature data, determine a training set and a test set based on the multi-dimensional feature data, train and test an initial sewing time prediction model by using the training set and the test set to obtain a target sewing time prediction model, and determine a sewing time prediction result corresponding to a sewing time prediction request based on the target sewing time prediction model; the initial sewing time prediction model is a sewing time prediction model determined based on the worker skill matrix and a light gradient boosting machine.

9. An electronic device, comprising: The method comprises: a memory configured to save a computer program; a processor configured to execute the computer program to implement the sewing production time prediction method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, a memory configured to save a computer program, wherein the computer program is executed by a processor to implement the sewing production time prediction method according to any one of claims 1 to 7.

Images