Production workshop manufacturing execution management system and method

Abstract

The present application provides a production workshop manufacturing execution management system and method, which relates to the field of intelligent management. It performs production workshop manufacturing execution management through data configuration, work order and product management, work-in-process monitoring, material tracking, process and quality management, working time statistics, equipment and energy monitoring, kanban display and barcode processing. Among them, the statistical work time of the workstation requires obtaining the sequence coding of the historical working time data and extracting the workstation characteristics as query features, so as to judge whether there is an abnormality in the working time according to the query feature and the query response guided by the range ratio between the implicit coding features of the working time sequence of each library workstation. In this way, a personalized normal working time model can be established, the working time abnormality can be accurately judged, and false alarms and missed alarms can be reduced. In addition, it is possible to systematically analyze historical data and mine trend patterns to provide early warning of potential problems and support preventive maintenance.

Description

Technical Field

[0001] The present application relates to the field of intelligent management, and more particularly, in an embodiment of the present application, to a production workshop manufacturing execution management system and method. Background Art

[0002] As the manufacturing industry advances towards intelligent and digital transformation, traditional manufacturing execution systems (MES) face many challenges. In modern production workshops, the production process is complex and changeable, involving multiple workstations, multiple equipment and a large amount of material flow. How to efficiently manage these resources and ensure product quality has become a key issue.

[0003] In this regard, patent CN117151625A discloses a workshop production and manufacturing execution management method and system, which covers functional modules such as data configuration, work order and product management, work-in-progress monitoring, material tracking, process and quality management, working time statistics, equipment and energy monitoring, kanban display, barcode processing, system settings and log recording. It realizes comprehensive digital management from production planning to execution.

[0004] This patent counts the working hours of each workstation and issues an alarm when abnormal working hours occur at the workstation. When the working hours of a workstation exceed the set working hours range, it is judged as abnormal working hours. However, the set working hours range is often static. In actual production, the reasons for working hours fluctuations are complex. If it is judged only by a fixed range, normal fluctuations are easily misjudged as abnormal, and abnormal working hours caused by special circumstances are easily judged with lags, affecting production efficiency and stability. In addition, traditional methods usually only focus on whether the current working hours exceed the preset range, while ignoring a large amount of valuable historical data, and failing to dig out deeper patterns and trends, thereby limiting the ability to provide early warning of potential problems.

[0005] Therefore, an optimized manufacturing execution management solution for production workshops is desired. Summary of the invention

[0006] In order to solve the above technical problems, the present application is proposed. The embodiment of the present application provides a production workshop manufacturing execution management system and method, which performs production workshop manufacturing execution management through data configuration, work order and product management, work-in-process monitoring, material tracking, process and quality management, working time statistics, equipment and energy monitoring, kanban display and barcode processing. Among them, it is necessary to obtain the sequence coding of historical working time data and extract the working station characteristics as query features to calculate the working time. The query response representation guided by the selection range ratio between the query features and the implicit coding features of the working time sequence of each library work station is used to judge whether there is an abnormality in the working time. In this way, a personalized normal working time model can be established, the working time abnormality can be accurately judged, and false alarms and missed reports can be reduced. In addition, it is possible to systematically analyze historical data and mine trend patterns to warn of potential problems in advance and support preventive maintenance.

[0007] According to one aspect of the present application, a manufacturing execution management method for a production workshop is provided, which includes: configuring system data, importing production work orders and issuing tasks; monitoring the task flow in the production workshop; monitoring the status parameters, process progress and workstation working hours of products in progress, and performing quality inspection on completed products; managing material consumption, batch traceability, line-side warehouses and inventory; counting workstation working hours and alarming for abnormal working hour data; monitoring equipment status and energy usage and analyzing the overall efficiency of the equipment; generating work order documents and managing finished product coding; wherein, counting workstation working hours and alarming for abnormal working hour data includes: obtaining historical working hour data of each workstation; serially encoding the historical working hour data of each workstation to obtain the workstation working hour time series A set of implicitly coded feature vectors; extracting the workstation working time sequence implicitly coded feature vector corresponding to the first workstation from the set of workstation working time sequence implicitly coded feature vectors as the query feature vector, and taking other workstation working time sequence implicitly coded feature vectors in the set of workstation working time sequence implicitly coded feature vectors as library workstation working time sequence implicitly coded feature vectors to obtain a set of library workstation working time sequence implicitly coded feature vectors; performing a selection range ratio-guided workstation working time feature invisible query response on the query feature vector and the set of library workstation working time sequence implicitly coded feature vectors to obtain a workstation working time characteristic query response coding vector; based on the workstation working time characteristic query response coding vector, obtaining a detection result for indicating whether there is an abnormality in the working time of the first workstation.

[0008] According to another aspect of the present application, a production workshop manufacturing execution management system is provided, which includes: a workstation historical working time data acquisition module, which is used to acquire the historical working time data of each workstation; a workstation historical working time sequence encoding module, which is used to sequence the historical working time data of each workstation to obtain a set of workstation working time sequence implicit coding feature vectors; a library workstation working time sequence implicit collection module, which is used to extract the workstation working time sequence implicit coding feature vector corresponding to the first workstation from the set of workstation working time sequence implicit coding feature vectors as a query feature vector, and ... The implicitly coded feature vectors of the working time sequence of other workstations are used as the implicitly coded feature vectors of the working time sequence of the library workstation to obtain a set of implicitly coded feature vectors of the working time sequence of the library workstation; a selection range ratio guided workstation working time feature invisible query response module is used to select the query feature vector and the set of implicitly coded feature vectors of the working time sequence of the library workstation to perform a selection range ratio guided workstation working time feature invisible query response to obtain a workstation working time characteristic query response coding vector; a work time abnormality detection result judgment module is used to obtain a detection result indicating whether there is an abnormality in the working time of the first workstation based on the work station working time characteristic query response coding vector.

[0009] Compared with the prior art, the present application provides a production workshop manufacturing execution management system and method, which performs production workshop manufacturing execution management through data configuration, work order and product management, work-in-process monitoring, material tracking, process and quality management, working time statistics, equipment and energy monitoring, kanban display and barcode processing. Among them, the statistical work time of the workstation requires obtaining the sequence coding of the historical working time data and extracting the workstation characteristics as query features, so as to judge whether there is an abnormality in the working time according to the query feature and the query response guided by the range ratio between the implicit coding features of the working time sequence of each library workstation to determine whether there is an abnormality in the working time. In this way, a personalized normal working time model can be established, the working time abnormality can be accurately judged, and false alarms and missed alarms can be reduced. In addition, it is possible to systematically analyze historical data and mine trend patterns to provide early warning of potential problems and support preventive maintenance. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0011] Figure 1 It is a flowchart of a production workshop manufacturing execution management method according to an embodiment of the present application.

[0012] Figure 2 The present invention is a flowchart for counting workstation working hours and issuing an alarm for abnormal working hour data in a production workshop manufacturing execution management method according to an embodiment of the present application.

[0013] Figure 3 This is a data flow diagram for counting workstation working hours and issuing an alarm for abnormal working hour data in a production workshop manufacturing execution management method according to an embodiment of the present application.

[0014] Figure 4 A flowchart of selecting a range ratio-guided workstation working time feature invisible query response for a set of the query feature vector and the library workstation working time sequence implicit coding feature vector in a production workshop manufacturing execution management method according to an embodiment of the present application to obtain a workstation working time feature query response coding vector.

[0015] Figure 5 In the production workshop manufacturing execution management method according to an embodiment of the present application, the workstation working time characteristic query response encoding vector is input into the classifier-based abnormal working time detection module to obtain the detection result, and the detection result is used to indicate whether there is an abnormality in the working time of the first workstation.

[0016] Figure 6 This is a system block diagram of a production workshop manufacturing execution management system according to an embodiment of the present application. DETAILED DESCRIPTION

[0017] Various exemplary embodiments, features and aspects of the present application will be described in detail below with reference to the accompanying drawings. The same reference numerals in the accompanying drawings represent elements with the same or similar functions. Although various aspects of the embodiments are shown in the accompanying drawings, the drawings are not necessarily drawn to scale unless otherwise specified.

[0018] The word “exemplary” is used exclusively herein to mean “serving as an example, example, or illustration.” Any embodiment described herein as “exemplary” is not necessarily to be construed as preferred or advantageous over other embodiments.

[0019] In addition, in order to better illustrate the present application, numerous specific details are given in the following specific embodiments. It should be understood by those skilled in the art that the present application can also be implemented without certain specific details. In some examples, methods, means, components and circuits well known to those skilled in the art are not described in detail in order to highlight the subject matter of the present application.

[0020] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the features. In the description of this application, the meaning of "plurality" is two or more, unless otherwise clearly and specifically defined.

[0021] As the manufacturing industry transforms toward intelligence, traditional manufacturing execution systems (MES) face challenges in managing complex production processes. Patent CN117151625A proposes a workshop production management method that covers work order management, work-in-process monitoring, material tracking, quality management, labor time statistics, equipment monitoring and other functions, realizing comprehensive digital management from production planning to execution. However, the judgment of abnormal working hours in traditional methods relies on a fixed range, which can easily misjudge normal fluctuations or delay the identification of abnormalities in special circumstances, thereby affecting production efficiency. At the same time, the failure to effectively use historical data to mine potential problems in production limits the ability to provide early warnings.

[0022] In response to the above technical problems, this application proposes a production workshop manufacturing execution management method. Figure 1 Flow chart of the manufacturing execution management method of a production workshop according to an embodiment of the present application. Figure 1As shown, the manufacturing execution management method for a production workshop according to an embodiment of the present application includes: step S110, configuring system data, importing production work orders and issuing tasks; step S120, monitoring the task flow of the production workshop; step S130, monitoring the status parameters, process progress and workstation working hours of the products in process, and performing quality inspection on the finished products; step S140, managing the consumption, batch traceability, line warehouse and inventory of materials; step S150, counting the workstation working hours and alarming for abnormal working hour data; step S160, monitoring the equipment status and energy usage and analyzing the comprehensive efficiency of the equipment; step S170, generating work order documents and managing finished product codes.

[0023] In the above-mentioned production workshop manufacturing execution management method, the step S110 configures system data, imports production work orders and issues tasks. It should be understood that configuring system data involves initializing all necessary parameters in the system, including but not limited to the specification parameters of production equipment, process routes, bill of materials (BOM), and standard working hours. Through accurate data configuration, each subsequent link can be operated based on accurate information, thereby reducing errors and improving overall work efficiency. Next, the production work order serves as a bridge connecting the production plan and actual production. It contains specific production task details, such as product model, quantity, delivery date and other key information. During the import process, the order information in the ERP (enterprise resource planning) system or other management system needs to be seamlessly connected to the MES (manufacturing execution system). This involves data format conversion, consistency checking and error handling mechanisms to ensure that each production work order can enter the MES system accurately. In addition, it is also necessary to consider the special requirements that may exist for different customers or orders, such as customized options or special quality standards, which need to be clearly marked in the work order and ensure that the MES can correctly interpret and execute them. When the production work order is successfully imported, the task instruction is issued. At this stage, MES will intelligently allocate tasks based on the current status of the workshop (such as equipment availability, personnel arrangements, etc.) and the specific conditions of each workstation. This means that we must not only consider how to most effectively use existing resources to achieve production goals, but also take into account possible changes or emergencies. That is, if a piece of equipment suddenly fails, MES should be able to quickly adjust the task allocation plan, re-plan the task sequence of the affected workstations, or find alternative equipment to minimize the impact on the overall production schedule.

[0024] In the above-mentioned production workshop manufacturing execution management method, the step S120 monitors the task flow of the production workshop. It should be understood that monitoring the task flow of the production workshop is a key link to ensure smooth production processes, improve production efficiency and product quality. Specifically, in a specific embodiment of the present application, the MES system uses advanced visualization technology to convert various collected data into intuitive and easy-to-understand charts and reports to help managers understand the task flow status of each workstation in the workshop at any time. Among them, through the dynamic Gantt chart, you can clearly see the progress of each work order at different workstations, which processes have been completed, which are still in progress, and when they are expected to be completed. At the same time, the system also sets up an early warning mechanism. Once an abnormal situation is found in a certain workstation (such as stagnation for a long time or exceeding the scheduled time), an alarm will be immediately issued to notify relevant personnel to deal with it in time. This instant feedback mechanism helps to respond quickly to sudden problems and reduce losses caused by delays. In addition, through the powerful data analysis tools equipped with the MES system, historical data can be mined in multiple dimensions to find potential trends and patterns. In the technical solution of the present application, by comparing and analyzing the production cycles of similar products in the past few months or even years, the key factors affecting production efficiency can be found out, and the production process can be adjusted or resource allocation can be optimized accordingly.

[0025] In the above-mentioned production workshop manufacturing execution management method, the step S130 monitors the state parameters, process progress and workstation working hours of the work-in-progress, and performs quality inspection on the finished products. It should be understood that, first, in a specific embodiment of the present application, for the monitoring of the state parameters of the work-in-progress, the MES system will combine historical data and current production conditions to establish a prediction model. The model can predict the state change trend of the work-in-progress based on the real-time collected data, identify potential problems in advance and take preventive measures. Among them, if the temperature of a certain process exceeds the normal range, the system can not only issue an alarm, but also suggest adjusting the operating parameters of the cooling system to ensure that the production environment remains stable. In addition, by analyzing the state parameters of different batches of products, the key factors affecting product quality can be found, the production process can be optimized, and the overall production efficiency can be improved. Further, in a specific embodiment of the present application, for tracking the progress of the process, the MES system obtains detailed production plan information through seamless docking with the ERP system, and decomposes it into specific tasks and assigns it to each workstation. After completing the corresponding task, the operator of each workstation updates the progress information in the system by scanning the device. In this way, the management can grasp the progress of each process in real time, find bottleneck links in time and make adjustments. For example, if the completion time of a certain process lags significantly behind the plan, the system will automatically prompt the manager and provide possible cause analysis and solutions. Furthermore, for the monitoring of workstation working hours, the MES system not only records the actual operation time of each workstation, but also compares it with the preset standard working hours to generate a detailed working hour report. Through the long-term accumulation and analysis of these data, enterprises can continuously optimize the process route, reasonably arrange human resources, and improve production efficiency. For example, if the actual working hours of a certain workstation are continuously higher than the standard working hours, it may mean that there is a bottleneck problem in the workstation, and further investigation of the reasons, such as aging equipment, insufficient operating skills, etc., is required, and corresponding improvement measures should be taken. Then, when the production task is completed, the next step is to conduct quality inspection of the finished product. The MES system is usually closely integrated with the quality management system (QMS) to ensure that every product can undergo a strict inspection procedure. In the technical solution of this application, the product is comprehensively inspected by automated inspection equipment (such as visual inspection systems, X-ray detectors, etc.) to capture surface defects, internal structural abnormalities and other problems. These inspection equipment can not only provide high-precision results, but also automatically generate detailed inspection reports through image processing technology. During the entire quality inspection process, the MES system also provides powerful data analysis functions. By mining and analyzing large amounts of quality inspection data, we can identify common quality problems and their root causes and develop targeted improvement measures.

[0026] In the above-mentioned production workshop manufacturing execution management method, the step S140 manages the consumption of materials, batch traceability, line-side warehouses and inventory. It should be understood that in a specific implementation of the present application, by attaching a QR code or RFID tag to each raw material and component, the time point of its entry and exit is recorded in real time to understand the consumption of materials. For example, before each process starts, the QR code or RFID tag on the material is scanned to record its entry time; similarly, after using these materials, scan again to record the actual consumption. In addition, weighing sensors, flow meters and other equipment can also be used to monitor the actual usage of materials to ensure the accuracy of the data. These data are transmitted to the MES system in real time via a wireless network to provide basic data support for subsequent analysis. By managing the consumption of materials, you can see the material usage of each work order at different workstations, which materials have been consumed, which are still in use, and when it is expected to be replenished. At the same time, the system also sets an early warning mechanism. Once it is found that the inventory of a certain material is lower than the safety inventory level, an alarm is immediately issued to notify relevant personnel to replenish it in time. This instant feedback mechanism helps to respond quickly to sudden problems and reduce downtime losses caused by material shortages. Further, in the technical solution of the present application, for material batch traceability, the MES system will combine historical data and current production conditions to establish a detailed batch tracking model. The model can track the source, usage and final destination of each material based on real-time collected data, ensuring that problems can be quickly located and appropriate measures can be taken when they occur. For example, if a product has a quality problem, the system can quickly find all products that use the same batch of raw materials based on the batch traceability function, and conduct targeted recall or rework. In addition, by analyzing the quality data of different batches of products, the key factors affecting product quality can be found, the production process can be optimized, and the overall production efficiency can be improved. Furthermore, for line-side warehouses and inventory management, the MES system obtains detailed inventory information through seamless docking with the ERP system, and decomposes it into specific tasks and assigns it to each warehouse manager. After completing the corresponding task, each warehouse manager updates the inventory information in the system by scanning the device. In this way, the management can grasp the inventory status of each material in real time, find bottlenecks in time and make adjustments. In this way, the MES system can also automatically generate replenishment requirements according to the production plan to ensure that the inventory is always maintained at a reasonable level. For the management of lineside warehouses, the MES system can not only monitor inventory levels in real time, but also optimize the placement and flow paths of materials. For example, by analyzing the frequency and amount of material requirements at different workstations, the system can recommend placing high-frequency materials closer to the production line to reduce handling time and costs.

[0027] In the above-mentioned production workshop manufacturing execution management method, the step S150 counts the work hours of the workstations and alarms for abnormal work hour data. The technical concept of the present application is to obtain the historical work hour data of each workstation, adapt the data analysis and coding method based on deep learning to sequence the historical work hour data of each workstation, and then extract the features corresponding to the first workstation from the implicit coding features of the work hour sequence of each workstation as the query feature, and use the remaining features as a set of library features, so as to intelligently judge whether there is an abnormality in the work hours of the first workstation according to the query feature and the query response guided by the selection range ratio between the implicit coding features of the work hour sequence of each library workstation. The present application can establish a personalized normal working time model for each workstation. In order to more accurately reflect the actual needs, the possibility of false alarms and omissions is reduced. In addition, the historical data can be systematically analyzed to fully explore the patterns and trends in the historical data to discover and warn potential problems earlier and support preventive maintenance.

[0028] Figure 2 The present invention is a flowchart for counting workstation working hours and issuing an alarm for abnormal working hour data in a production workshop manufacturing execution management method according to an embodiment of the present application. Figure 3 The following is a data flow diagram for counting workstation working hours and issuing an alarm for abnormal working hours data in a production workshop manufacturing execution management method according to an embodiment of the present application. Figure 2 and Figure 3 As shown, in the above-mentioned production workshop manufacturing execution management method, the step S150, counting the work time of the workstation and alarming for abnormal work time data, includes: S151, obtaining the historical work time data of each workstation; S152, serially encoding the historical work time data of each workstation to obtain a set of implicit coding feature vectors of the workstation work time sequence; S153, extracting the workstation work time sequence implicit coding feature vector corresponding to the first workstation from the set of implicit coding feature vectors of the workstation work time sequence as the query feature vector, and the workstation work time sequence implicit coding feature vector is used as the query feature vector. The implicitly coded feature vectors of the working time sequence of other workstations in the set of feature vectors are used as the implicitly coded feature vectors of the working time sequence of the library workstation to obtain a set of implicitly coded feature vectors of the working time sequence of the library workstation; S154, a workstation working time feature invisible query response guided by a selection range ratio is performed on the query feature vector and the set of implicitly coded feature vectors of the working time sequence of the library workstation to obtain a workstation working time feature query response coding vector; S155, based on the workstation working time feature query response coding vector, a detection result indicating whether there is an abnormality in the working time of the first workstation is obtained.

[0029] Specifically, in step S151, the historical working hours data of each workstation is obtained. It should be understood that the working hours of each workstation will be affected by many factors, such as equipment status, raw material quality, operator skills, etc. Relying only on a fixed range for judgment, it is easy to misjudge normal fluctuations as abnormalities, or to lag in the judgment of abnormal working hours caused by special circumstances. In this way, the traditional static setting method is not flexible and accurate enough for optimization in the face of complex production environments. By introducing historical working hours data, the actual working mode and fluctuation of each workstation can be captured, so as to more scientifically evaluate whether the current working hours are normal, and take measures in advance to avoid similar problems from happening again.

[0030] In an embodiment of the present application, the step S152, sequence encoding the historical working time data of each workstation to obtain a set of implicitly encoded feature vectors of the workstation working time sequence, includes: using a sequence encoder to encode the historical working time data of each workstation to obtain a set of implicitly encoded feature vectors of the workstation working time sequence. More specifically, in an embodiment of the present application, the sequence encoder is a sequence encoder based on a forward LSTM model. It should be understood that considering that the working time data has the characteristics of a time series, that is, there is a temporal sequence and dependency relationship between data points. Therefore, in order to effectively capture and represent this time series information, such as the changing trend and periodic law of the working time of a certain workstation over a period of time, the present application obtains a set of implicitly encoded feature vectors of the workstation working time sequence by sequence encoding the historical working time data of each workstation. In particular, in an example of the present application, a sequence encoder based on a forward LSTM model is used to encode the historical working time data of each workstation to obtain a set of implicitly encoded feature vectors of the workstation working time sequence. It should be understood that historical working hours data is a typical time series data with the characteristics of front-end and back-end dependencies. The forward LSTM model can handle this kind of sequence data well. Through its internal memory unit, it can effectively capture and utilize long-term time dependencies, accurately model the time series information in the working hours data, learn these complex patterns, effectively extract various implicit features in the working hours data, and form a more representative time series implicit coding feature vector, providing richer information for subsequent analysis and decision-making.

[0031] Specifically, in step S153, the workstation work time sequence implicit coding feature vector corresponding to the first workstation is extracted from the set of workstation work time sequence implicit coding feature vectors as the query feature vector, and other workstation work time sequence implicit coding feature vectors in the set of workstation work time sequence implicit coding feature vectors are used as library workstation work time sequence implicit coding feature vectors to obtain a set of library workstation work time sequence implicit coding feature vectors. It should be understood that, considering that each workstation has different characteristics and tasks, in order to conduct a detailed analysis of the specific workstation, capture and mine whether there is an abnormality between the workstation working hours and other work, the present application extracts the workstation working hours sequence implicit coding feature vector corresponding to the first workstation from the set of the workstation working hours sequence implicit coding feature vector as the query feature vector, and uses the other workstation working hours sequence implicit coding feature vectors in the set of the workstation working hours sequence implicit coding feature vector as the library workstation working hours sequence implicit coding feature vector to obtain the set of library workstation working hours sequence implicit coding feature vectors. In this way, the first workstation feature vector is used as the query feature vector and compared with the set of library workstation working hours sequence implicit coding feature vectors. If the first workstation working hours are abnormal, its feature vector will be significantly different from other vectors in the set. By analyzing this difference, it can be intelligently judged whether the first workstation working hours are abnormal. For example, under normal circumstances, the working time characteristic vectors of each workstation fluctuate within a certain range. If the characteristic vector of the first workstation deviates from this range, it may indicate abnormal working time, thereby enabling more detailed and focused analysis and processing of a specific workstation situation.

[0032] Figure 4 The flowchart of selecting the range ratio-guided invisible query response of the workstation working time characteristic to obtain the workstation working time characteristic query response encoding vector in the production workshop manufacturing execution management method according to the embodiment of the present application. Figure 4As shown, in an embodiment of the present application, the step S154, which performs a selection range ratio-guided invisible query response of the station working time feature on the query feature vector and the set of implicitly coded feature vectors of the library station working time sequence to obtain a station working time feature query response coding vector, includes: S1541, based on the query feature vector, determining a query-library station working time sequence dynamic search anchor vector from the set of implicitly coded feature vectors of the library station working time sequence; S1542, based on the feature distribution characteristics of the query-library station working time sequence dynamic search anchor vector, determining a library station working time sequence selection range ratio; S1543, based on the query-library station working time sequence dynamic search anchor vector, performing a dynamic semantic feature query response on each feature in the query feature vector and the library station working time sequence selection range ratio to obtain the station working time feature query response coding vector. It should be understood that in the step S154, it is considered that there is a certain semantic association relationship between the query feature vector and the set of implicitly coded feature vectors of the workstation working time sequence in the library. Therefore, in order to fully reflect the relationship between the workstation working time and other workstations, provide a clear and accurate basis for in-depth understanding of the working time characteristics, and efficiently query and match and deeply understand the information in these large-scale feature vector libraries, the present application selects the range ratio-guided workstation working time feature invisible query response for the query feature vector and the set of implicitly coded feature vectors of the workstation working time sequence in the library to obtain the workstation working time feature query response coding vector. In particular, by selecting the range ratio-guided method, it is possible to identify the feature vector that best represents the query intention from the query feature vector and the set of implicitly coded feature vectors of the workstation working time sequence in the library, and accurately locate the features that are important for understanding the law of changes in workstation working time and judging work time anomalies.

[0033] In an embodiment of the present application, the step S1541, based on the query feature vector, determines the query-warehouse work station working time sequence dynamic search anchor vector from the set of the library work station working time sequence implicit coding feature vectors, including: S1541-1, calculating the internal relationship score value of the query feature vector relative to each library work station working time sequence implicit coding feature vector in the set of the library work station working time sequence implicit coding feature vectors to obtain a set of query-warehouse work station working time sequence implicit coding feature internal relationship score values; S1541-2, using the library work station working time sequence implicit coding feature vector corresponding to the largest query-warehouse work station working time sequence implicit coding feature internal relationship score value in the set of the query-warehouse work station working time sequence implicit coding feature internal relationship score values ​​as the query-warehouse work station working time sequence dynamic search anchor vector.

[0034] Specifically, in step S1541-1, the internal relationship score of the query feature vector relative to each implicit coding feature vector in the set of implicit coding feature vectors of the warehouse work station work time sequence is calculated to obtain a set of query-warehouse work station work time sequence implicit coding feature internal relationship score values, which is expressed as the internal relationship score formula: ;in, is the set of implicitly encoded feature vectors of the working time sequence of the warehouse workstations, and is the first, second, and third in the set of implicitly encoded feature vectors of the warehouse workstation working time sequence. and The implicit encoding feature vector of the time series of the work time of each warehouse station, is the weight matrix of the warehouse workstation hours, is the query feature vector, is the query weight matrix, is the hyperbolic tangent function, is the weight coefficient vector, It is the first in the set of internal relationship scores of query-warehouse workstation working time sequence implicit coding feature query-warehouse workstation working time sequence implicit coding feature The internal relationship score of the query-warehouse workstation working time sequence implicit coding feature query-warehouse workstation working time sequence implicit coding feature. It should be understood that the internal relationship score of the query-warehouse workstation working time sequence implicit coding feature not only reflects the direct similarity between the query feature vector and the warehouse workstation working time sequence implicit coding feature vector, but also implies the potential semantic connection. By comparing the internal relationship score values ​​of the two, the similarity and correlation between the query feature vector and each warehouse workstation working time sequence implicit coding feature vector can be quantified, thereby determining the most relevant features of the query feature vector. This helps to narrow the scope of the data set for subsequent processing and improve query efficiency and accuracy.

[0035] Specifically, in step S1541-2, the implicit coding feature vector of the warehouse work time sequence corresponding to the largest query-warehouse work time sequence implicit coding feature internal relationship score value in the set of query-warehouse work time sequence implicit coding feature internal relationship score values ​​is used as the query-warehouse work time sequence dynamic search anchor vector, and the query-warehouse work time sequence dynamic search formula is expressed as: ;in, is the vector at the location of the maximum value, It is the query-library workstation working time sequence query-library workstation working time sequence dynamic search anchor vector. It should be understood that this step is not just to find the single point with the highest score, but to comprehensively consider the global distribution characteristics and ensure that the selected anchor point is representative. The selected query-library workstation working time sequence dynamic search anchor vector not only reflects the central tendency of the query, but also serves as the basis for further optimization. Among them, the feature vector with the highest score is selected as the anchor point, which means that it has the strongest correlation or similarity with the query feature vector. This not only simplifies the subsequent search process, but also increases the possibility of finding the optimal solution, thereby guiding a more precise search direction and scope.

[0036] In an embodiment of the present application, the step S1542 determines the library station work time sequence selection range ratio based on the characteristic distribution characteristics of the query-library work time sequence dynamic search anchor vector, including: S1542-1, calculating the size of the library station work time sequence selection range ratio based on the mean, variance and maximum value of the query-library work time sequence dynamic search anchor quantity; S1542-2, based on the size of the library station work time sequence selection range ratio, anchoring the set of library station work time sequence implicit coding feature vectors to obtain the library station work time sequence selection range ratio, wherein the vector of the starting position of the library station work time sequence selection range ratio is the query-library work time sequence dynamic search anchor vector, and each library station work time sequence implicit coding feature vector in the library station work time sequence selection range ratio is defined as a library station work time sequence implicit coding feature dynamic search optimization feature vector.

[0037] In step S1542-1, the size of the selection range ratio of the library work station time sequence is calculated based on the mean, variance and maximum value of the query-library work station time sequence dynamic search anchor quantity, and the size formula of the selection range ratio of the library work station time sequence is expressed as: ;in, and They are The variance and mean of is the modulation parameter, To obtain The maximum value in Indicates rounding up. It is the size of the selection range ratio of the library workstation time sequence. It should be understood that according to the selected query-library workstation time sequence dynamic search anchor vector and its statistical characteristics (mean, variance, maximum value), a selection range ratio is calculated to limit the search range, which can help the system focus more on the area where the potential best answer is located, thereby effectively reducing unnecessary calculations and improving search efficiency and accuracy. At the same time, it can ensure that the possible optimal solution is covered. This strategy balances the breadth and depth of the search, making the whole process efficient and comprehensive.

[0038] In step S1542-2, based on the size of the library station work time sequence selection range ratio, the set of the library station work time sequence implicit coding feature vectors is anchored to obtain the library station work time sequence selection range ratio, wherein the vector of the starting position of the library station work time sequence selection range ratio is the query-library station work time sequence dynamic search anchor vector, and each library station work time sequence implicit coding feature vector in the library station work time sequence selection range ratio is defined as a library station work time sequence implicit coding feature dynamic search optimization feature vector, which is expressed as an anchor formula: ;in, It is a set of dynamically searched optimized feature vectors of implicit coding features of the library station working time sequence in the library station working time sequence selection range ratio. It should be understood that this step is intended to focus on those feature vectors that have a high correlation with the query, while excluding irrelevant parts, so as to narrow the search space, focus on the most potential candidates, and improve efficiency. Among them, the set of feature vectors after screening is smaller in scale and contains more elements that are highly correlated with the query feature vector. This not only speeds up the search, but also improves the quality of the results. Here, the determination of the selection range ratio is usually combined with statistical analysis or machine learning models to ensure that the selected optimized library station working time sequence implicit coding feature vector can retain the original information to the greatest extent and reduce noise interference. It is worth noting that as new data continues to flow in, the selection range and the optimized library station working time sequence implicit coding feature vectors within it should be dynamically updated to maintain the timeliness and relevance of the system.

[0039] Specifically, in step S1543, based on the query-warehouse workstation working time sequence dynamic search anchor vector, the query feature vector and each feature in the warehouse workstation working time sequence selection range ratio are subjected to dynamic semantic feature query response to obtain the workstation working time feature query response encoding vector, which is expressed as a dynamic semantic feature query response formula: ;in, and The first and The implicit coding features of the time sequence of each warehouse workstation are dynamically searched and optimized for feature vectors. is the Euclidean norm of the vector, for and The semantic similarity between is the workstation work time characteristic query response encoding vector. It should be understood that the query-library workstation work time sequence dynamic search anchor vector and each dynamic search optimization library workstation work time sequence implicit encoding feature vector in the selection range ratio are input into the dynamic semantic search encoder to obtain the workstation work time characteristic query response encoding vector, wherein the dynamic semantic search encoder itself is a complex neural network architecture, and its design goal is an embedding space that can effectively express semantic information. In this space, samples originally scattered in the high-dimensional feature space are reorganized into a compact representation with clear semantic meaning. The final output workstation work time characteristic query response encoding vector has a high degree of abstract generalization ability and can show excellent performance in various downstream tasks. This process not only emphasizes capturing the macro-semantic similarity between the two, but also pays special attention to the subtle but critical difference information therein. Although the query feature vector and the dynamic semantic feature query response of each feature in the library workstation work time sequence selection range ratio may be highly similar in overall semantics, the subtle differences between them are crucial for subsequent processing. For example, in complex production scheduling or resource management scenarios, even small time deviations or resource configuration differences may lead to completely different results. Therefore, methods such as cosine similarity are used to quantify and enhance these subtle differences, ensuring that no important details are overlooked during the analysis process. Specifically, cosine similarity, as a method of measuring the angle between two non-zero vectors, can effectively distinguish feature vectors that are very close but not exactly the same even in high-dimensional space. It not only captures the overall semantic association between the query feature vector and each feature vector in the database, but also highlights the subtle differences that are critical to the final decision. This method compares the values ​​on each feature dimension and calculates a score between 0 and 1. The closer the score is to 1, the more consistent the direction of the two is, that is, the more similar the semantics is; while the closer it is to 0, the greater the difference between the two. In this step, this technology can not only identify the most relevant feature vectors, but also carefully distinguish the subtle differences between different feature vectors, thus providing a solid foundation for subsequent precise matching and optimization. In this way, the system is able to generate a detailed encoding vector for the query response to the workstation work time characteristics, which accurately reflects the query requirements of the workstation work time characteristics and provides an answer that meets expectations.

[0040] In an embodiment of the present application, the step S155, based on the workstation working time characteristic query response coding vector, obtains a detection result for indicating whether the working time of the first workstation is abnormal, including: inputting the workstation working time characteristic query response coding vector into the abnormal working time detection module based on the classifier to obtain the detection result, and the detection result is used to indicate whether the working time of the first workstation is abnormal. That is, the workstation working time characteristic query response coding vector obtained by querying the response using the set of the query feature vector and the implicit coding feature vector of the library workstation working time sequence is classified and processed, so as to utilize the powerful ability of the classifier in pattern recognition and data classification to learn the complex information contained in the workstation working time characteristic query response coding vector and identify the pattern therein, and accurately classify it as normal or abnormal. This provides clear information for production management personnel, so that they can timely understand the abnormal fluctuations of working time in the production process, so as to take corresponding measures to adjust and optimize.

[0041] Figure 5 In the manufacturing execution management method of the production workshop according to the embodiment of the present application, the workstation working time characteristic query response encoding vector is input into the abnormal working time detection module based on the classifier to obtain the detection result, and the detection result is used to indicate whether the working time of the first workstation is abnormal. Figure 5 As shown, in an embodiment of the present application, in the step S155, the workstation working time characteristic query response coding vector is input into the classifier-based abnormal working time detection module to obtain the detection result, and the detection result is used to indicate whether the working time of the first workstation is abnormal, including: S1551, using the fully connected layer of the classifier-based abnormal working time detection module to fully connect the workstation working time characteristic query response coding vector to obtain a fully connected coding feature vector of the workstation working time characteristic; S1552, passing the fully connected coding feature vector of the workstation working time characteristic through the Softmax classification function of the classifier-based abnormal working time detection module to obtain a first probability that the working time belonging to the first workstation is abnormal and a second probability that the working time belonging to the first workstation is not abnormal; S1553, based on the comparison between the first probability and the second probability, determining the detection result. In this way, subtle changes and potential problems in working time data can be accurately captured, and a quantitative and reliable basis for decision-making can be provided to help managers promptly discover and deal with abnormal situations that may affect production efficiency or quality, so as to effectively improve production stability and reliability and reduce cost increases and efficiency decreases caused by abnormal working hours.

[0042] Preferably, in a specific embodiment of the present application, the workstation working time characteristic query response coding vector is input into an abnormal working time detection module based on a classifier to obtain a detection result, including: determining a workstation working time characteristic query response probability value obtained by inputting the workstation working time characteristic query response coding vector into the abnormal working time detection module based on a classifier, wherein the workstation working time characteristic query response probability value indicates the probability that the working time of the first workstation is abnormal; multiplying the feature mean of the workstation working time characteristic query response coding vector by the workstation working time characteristic query response probability value to obtain a workstation working time characteristic query response statistical field value, that is: ;in, represents the feature mean of the workstation working time characteristic query response encoding vector, represents the response probability value of the workstation working time characteristic query, Indicates the statistical field value of the workstation working time characteristic query response.

[0043] Subtract one from the statistical field value of the workstation working time characteristic query response and divide it by the statistical field value of the workstation working time characteristic query response to obtain the partial probability value of the workstation working time characteristic query response, that is: ;in, It represents the partial probability value of the response to the workstation working hours characteristic query.

[0044] Calculate the power function of the workstation working time characteristic query response encoding vector with the workstation working time characteristic query response partial probability value as the exponent, and multiply it by the workstation working time characteristic query response partial probability value to obtain the workstation working time characteristic query response microscopic representation vector, that is: ,in, represents the encoding vector of the query response to the workstation working hours feature, Represents the microscopic representation vector of the query response to the workstation working time characteristic, It means point multiplication by position. It represents a power function which calculates the encoding vector of the workstation working time characteristic query response with the partial probability value of the workstation working time characteristic query response as the exponent.

[0045] After multiplying the workstation working time characteristic query response encoding vector by the workstation working time characteristic query response partial probability value, an exponential function with a natural constant as the base is calculated to obtain the workstation working time characteristic query response macroscopic mapping vector, that is: ;in, represents an exponential function with a natural constant as base, Represents the macro mapping vector of the query response of the workstation working time characteristics.

[0046] After calculating the logarithm value of the microscopic representation vector of the workstation working time characteristic query response with base 2, it is weightedly summed with the macroscopic mapping vector of the workstation working time characteristic query response to obtain the optimized workstation working time characteristic query response encoding vector, that is: ;in, represents the first weight parameter, represents the second weight parameter, represents the logarithmic function with base 2, represents the optimized workstation working time characteristic query response encoding vector, Indicates addition by position.

[0047] The optimized workstation working time characteristic query response encoding vector is input into the classifier-based abnormal working time detection module to obtain the detection result.

[0048] Since the query feature vector and the set of implicit coding feature vectors of the library workstation working time sequence represent the sequence coding features of the historical working time data of the first workstation and the historical working time data of other workstations respectively, when performing the implicit query response of the workstation working time feature guided by the selection range ratio, the sequence coding features at different sample space scales will have a micro-macro representation deviation, which will cause a dynamic deviation in the mapping of features to class target probabilities when the workstation working time feature query response coding vector is input into the classifier-based abnormal working time detection module, thereby reducing the accuracy of the detection results.

[0049] Therefore, the partial low-order derivatives of the statistical distribution field corresponding to the workstation working time characteristic query response coding vector are used as non-overlapping macro-feature representation behavior patches of the workstation working time characteristic query response coding vector, so as to organize the space of different macro-behavior patches under the non-isotropic backbone structure of the workstation working time characteristic query response coding vector, so as to enhance the dynamic sensitivity of the long-series micro-complex information distribution of the workstation working time characteristic query response coding vector to the class probability macro-representation behavior, thereby promoting the iterative dynamic consistency of the class target between the classification target and the extracted features during the feature space-class probability mapping, so as to improve the accuracy of the detection result obtained by the abnormal working time detection module based on the classifier input of the workstation working time characteristic query response coding vector.

[0050] In summary, the step S150 is explained, which obtains the historical working time data of each workstation, adapts to the data analysis and coding method based on deep learning to sequence the historical working time data of each workstation, and then extracts the features corresponding to the first workstation from the implicit coding features of the working time sequence of each workstation as the query feature, and uses the remaining features as a set of library features, so as to intelligently judge whether there is an abnormality in the working time of the first workstation based on the query feature and the query response guided by the selection range ratio between the query feature and the implicit coding features of the working time sequence of each library workstation. The present application can establish a personalized normal working time model for each workstation. In order to more accurately reflect the actual needs, the possibility of false alarms and omissions is reduced. In addition, it can systematically analyze historical data and fully explore the patterns and trends in historical data to discover and warn potential problems earlier and support preventive maintenance.

[0051] In the above-mentioned production workshop manufacturing execution management method, the step S160 monitors the equipment status and energy usage and analyzes the overall efficiency of the equipment. It should be understood that in a specific embodiment of the present application, by installing a variety of sensors on the production equipment, such as temperature sensors, vibration sensors, pressure sensors, etc., to monitor the key parameters of the equipment in real time during operation, so as to monitor the equipment status. Further, for the monitoring of energy usage, the technical solution of the present application monitors the energy consumption on the production line in real time through the MES or SCADA system by installing smart meters, water meters, gas meters and other equipment. These meters can not only provide accurate energy consumption data, but also transmit data to the system in real time through wireless networks to provide basic data support. For example, the power consumption of each device can be monitored in real time through smart meters, and the energy consumption differences under different working modes can be understood; the water and gas consumption on the production line can be monitored through water meters and gas meters, and their impact on overall energy consumption can be evaluated. Furthermore, in the process of monitoring energy usage, the system will record and analyze the collected data in detail. By comparing energy consumption data in different time periods, high-energy consumption periods and low-energy consumption periods can be identified and corresponding energy-saving strategies can be formulated. For example, if it is found that the energy consumption of a certain device during non-peak hours is significantly higher than that of other devices, the system can recommend adjusting its operating time and moving it to the off-peak period to reduce electricity costs. After the data on equipment status and energy usage are fully collected and processed, the next step is to analyze the overall efficiency of the equipment. Overall efficiency usually includes multiple dimensions such as equipment utilization (OEE, Overall Equipment Effectiveness) and energy efficiency. MES or SCADA systems calculate these indicators to help management fully understand the operating status of the equipment. For example, equipment utilization can be measured by calculating the proportion of the actual operating time of the equipment to the total planned time; energy efficiency can be evaluated by calculating the energy consumption per unit of output. Through the long-term accumulation and analysis of these indicators, enterprises can continuously optimize process routes, reasonably arrange equipment maintenance, and improve production efficiency.

[0052] In the above-mentioned production workshop manufacturing execution management method, the step S170 generates work order documents and manages finished product codes. It should be understood that the generated work order documents are seamlessly connected through the ERP (Enterprise Resource Planning) system and the MES (Manufacturing Execution System) to ensure that the production plan information can be accurately transmitted to the production workshop. Specifically, when the sales department receives a customer order, the order information will be processed through the ERP system, and a detailed production plan will be formulated based on factors such as inventory status and production capacity. These plans are then broken down into specific work order tasks and assigned to each production workshop or workstation. In order to ensure the accuracy of the work order information, the ERP system will automatically generate detailed work order documents. Each work order document contains rich information, such as product model, quantity, delivery date, bill of materials (BOM), process route, etc. This information not only guides the operators of the production line on how to perform production operations, but also provides the necessary basis for subsequent quality inspections, finished product warehousing, and other links. All generated work order documents will be transmitted to the MES system for centralized processing and storage. In order to facilitate subsequent query and analysis, the system will archive and manage work orders according to different classification standards (such as date, work order number, product type, etc.). Furthermore, in addition to generating work order documents, it is also crucial to manage finished product codes. Finished product codes are not only unique identifiers for products, but also important tools throughout the entire production and logistics process. By assigning a unique code to each finished product, full traceability from production to delivery can be achieved. Finished product codes usually consist of a series of characters, including key information such as product model, batch number, and production date. These codes can not only help companies quickly locate and track products in internal management systems, but also provide customers with transparent product sources and quality assurance. Among them, when a batch of products is completed and passes quality inspection, the system will automatically generate the corresponding finished product code and associate it with all relevant information of the batch (such as raw material source, production process parameters, quality inspection reports, etc.). This information is not only stored in the MES system, but also uploaded to the company's database for other departments (such as sales department, customer service department, etc.) to consult and use. For example, when a customer makes an after-sales service request, the customer service staff can quickly query the production record and historical maintenance status of the product through the finished product code to provide more accurate service support. In order to ensure the uniqueness and traceability of the finished product code, each finished product code needs to be verified by the system to ensure that it meets the preset format requirements and remains unchanged throughout the production cycle. In addition, the system will regularly back up the coding data to prevent data loss due to unexpected failures or human operational errors. In this way, companies can access complete finished product information at any time and anywhere, providing strong support for decision-making.

[0053] In summary, the manufacturing execution management method of the production workshop based on the embodiment of the present application is explained, which performs manufacturing execution management of the production workshop through data configuration, work order and product management, work-in-process monitoring, material tracking, process and quality management, working time statistics, equipment and energy monitoring, kanban display and barcode processing. Among them, it is necessary to obtain the sequence coding of the historical working time data and extract the working station characteristics as query features to judge whether there is an abnormality in the working time according to the query feature and the query response guided by the range ratio between the query feature and the implicit coding features of the working time sequence of each library work station to determine whether there is an abnormality in the working time. In this way, a personalized normal working time model can be established to accurately judge the abnormality of working time and reduce false alarms and missed reports. In addition, it is possible to systematically analyze historical data and mine trend patterns to warn of potential problems in advance and support preventive maintenance.

[0054] Figure 6 1 is a system block diagram of a production workshop manufacturing execution management system according to an embodiment of the present application. Figure 6 As shown, according to the manufacturing execution management system 100 of the production workshop of the embodiment of the present application, it includes: a workstation historical working time data acquisition module 110, which is used to obtain the historical working time data of each workstation; a workstation historical working time sequence encoding module 120, which is used to sequence the historical working time data of each workstation to obtain a set of workstation working time sequence implicit coding feature vectors; a library workstation working time sequence implicit collection module 130, which is used to extract the workstation working time sequence implicit coding feature vector corresponding to the first workstation from the set of the workstation working time sequence implicit coding feature vectors as the query feature vector, and add the workstation working time sequence implicit coding feature vector in the set of the workstation working time sequence implicit coding feature vectors to the query feature vector. The implicit coding feature vectors of the working time sequence of other workstations are used as the implicit coding feature vectors of the working time sequence of the library workstation to obtain a set of implicit coding feature vectors of the working time sequence of the library workstation; a selection range ratio guided workstation working time feature invisible query response module 140 is used to select the query feature vector and the set of implicit coding feature vectors of the working time sequence of the library workstation to perform a selection range ratio guided workstation working time feature invisible query response to obtain a workstation working time characteristic query response coding vector; a work time abnormality detection result judgment module 150 is used to obtain a detection result indicating whether there is an abnormality in the working time of the first workstation based on the work station working time characteristic query response coding vector.

[0055] Here, those skilled in the art can understand that the specific operations of each step in the above-mentioned production workshop manufacturing execution management system have been referred to above. Figures 1 to 5 The invention has been described in detail in the description of the manufacturing execution management method for a production workshop, and therefore, its repeated description will be omitted.

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

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

[0058] In summary, the manufacturing execution management system of the production workshop based on the embodiment of the present application is explained, which performs manufacturing execution management of the production workshop through data configuration, work order and product management, work-in-process monitoring, material tracking, process and quality management, working time statistics, equipment and energy monitoring, kanban display and barcode processing. Among them, it is necessary to obtain the sequence coding of the historical working time data and extract the working station characteristics as query features to judge whether there is an abnormality in the working time according to the query feature and the query response guided by the range ratio between the query feature and the implicit coding features of the working time sequence of each library work station to determine whether there is an abnormality in the working time. In this way, a personalized normal working time model can be established to accurately judge the abnormality of working time and reduce false alarms and missed reports. In addition, it is possible to systematically analyze historical data and mine trend patterns to warn of potential problems in advance and support preventive maintenance.

Claims

1. A production workshop manufacturing execution management method, comprising: Configure system data, import production work orders and issue tasks; monitor the task flow in the production workshop; Monitor the status parameters, process progress and workstation working hours of products in process, and perform quality inspection on finished products; manage material consumption, batch traceability, line warehouse and inventory; count workstation working hours and issue alarms for abnormal working hour data; monitor equipment status and energy usage and analyze the overall efficiency of the equipment; generate work order documents and manage finished product coding; it is characterized in that the workstation working hours are counted and the abnormal working hour data are alarmed, including: obtaining the historical working hour data of each workstation; performing sequence coding on the historical working hour data of each workstation to obtain a set of implicit coding feature vectors of the workstation working hour sequence; from the set of implicit coding feature vectors of the workstation working hour sequence The workstation working time sequence implicit coding feature vector corresponding to the first workstation is extracted as the query feature vector, and other workstation working time sequence implicit coding feature vectors in the set of the workstation working time sequence implicit coding feature vectors are used as library workstation working time sequence implicit coding feature vectors to obtain a set of library workstation working time sequence implicit coding feature vectors; a range ratio-guided workstation working time feature invisible query response is performed on the query feature vector and the set of library workstation working time sequence implicit coding feature vectors to obtain a workstation working time feature query response coding vector; based on the workstation working time feature query response coding vector, a detection result indicating whether there is an abnormality in the working time of the first workstation is obtained.

2. The manufacturing execution management method for a production workshop according to claim 1, characterized in that: The historical working time data of each workstation is sequentially encoded to obtain a set of implicitly encoded feature vectors of the working time sequence of the workstation, including: using a sequence encoder to encode the historical working time data of each workstation to obtain a set of implicitly encoded feature vectors of the working time sequence of the workstation.

3. The manufacturing execution management method for a production workshop according to claim 2, characterized in that: The sequence encoder is a sequence encoder based on a forward LSTM model.

4. The manufacturing execution management method for a production workshop according to claim 3, characterized in that: The query feature vector and the set of implicitly coded feature vectors of the library workstation working time sequence are subjected to a selection range ratio-guided invisible query response of the workstation working time feature to obtain a workstation working time feature query response coding vector, including: based on the query feature vector, determining a query-library workstation working time sequence dynamic search anchor vector from the set of implicitly coded feature vectors of the library workstation working time sequence; based on the feature distribution characteristics of the query-library workstation working time sequence dynamic search anchor vector, determining a library workstation working time sequence selection range ratio; based on the query-library workstation working time sequence dynamic search anchor vector, performing a dynamic semantic feature query response on each feature in the query feature vector and the library workstation working time sequence selection range ratio to obtain the workstation working time feature query response coding vector.

5. The manufacturing execution management method for a production workshop according to claim 4, characterized in that: Based on the query feature vector, a query-warehouse work station working time sequence dynamic search anchor vector is determined from the set of the warehouse work station working time sequence implicit coding feature vectors, including: calculating the internal relationship score value of the query feature vector relative to each warehouse work station working time sequence implicit coding feature vector in the set of the warehouse work station working time sequence implicit coding feature vectors to obtain a set of query-warehouse work station working time sequence implicit coding feature internal relationship score values; using the warehouse work time sequence implicit coding feature vector corresponding to the largest query-warehouse work station working time sequence implicit coding feature internal relationship score value in the set of the query-warehouse work station working time sequence implicit coding feature internal relationship score values ​​as the query-warehouse work station working time sequence dynamic search anchor vector.

6. The manufacturing execution management method for a production workshop according to claim 5, characterized in that: Based on the characteristic distribution characteristics of the query-warehouse workstation work time sequence dynamic search anchor vector, determine the library workstation work time sequence selection range ratio, including: calculating the size of the library workstation work time sequence selection range ratio based on the mean, variance and maximum value of the query-warehouse workstation work time sequence dynamic search anchor quantity; based on the size of the library workstation work time sequence selection range ratio, anchor the set of library workstation work time sequence implicit coding feature vectors to obtain the library workstation work time sequence selection range ratio, wherein the vector of the starting position of the library workstation work time sequence selection range ratio is the query-warehouse workstation work time sequence dynamic search anchor vector, and each library workstation work time sequence implicit coding feature vector in the library work time sequence selection range ratio is defined as a library workstation work time sequence implicit coding feature dynamic search optimization feature vector.

7. The manufacturing execution management method for a production workshop according to claim 6, characterized in that: Based on the workstation working time characteristic query response coding vector, a detection result is obtained to indicate whether the working time of the first workstation is abnormal, including: inputting the workstation working time characteristic query response coding vector into a classifier-based abnormal working time detection module to obtain the detection result, and the detection result is used to indicate whether the working time of the first workstation is abnormal.

8. The manufacturing execution management method for a production workshop according to claim 7, characterized in that: The workstation working time characteristic query response encoding vector is input into the classifier-based abnormal working time detection module to obtain the detection result, and the detection result is used to indicate whether the working time of the first workstation is abnormal, including: using the fully connected layer of the classifier-based abnormal working time detection module to fully connect the workstation working time characteristic query response encoding vector to obtain a fully connected encoding feature vector of the workstation working time characteristic; passing the fully connected encoding feature vector of the workstation working time characteristic through the Softmax classification function of the classifier-based abnormal working time detection module to obtain a first probability that the working time attributable to the first workstation is abnormal and a second probability that the working time attributable to the first workstation is not abnormal; based on the comparison between the first probability and the second probability, determining the detection result.

9. A production workshop manufacturing execution management system, characterized in that: include: A workstation historical working time data acquisition module is used to acquire the historical working time data of each workstation; a workstation historical working time sequence encoding module is used to perform sequence encoding on the historical working time data of each workstation to obtain a set of implicit encoding feature vectors of the workstation working time sequence; A library workstation working time sequence implicit set module is used to extract the workstation working time sequence implicit coding feature vector corresponding to the first workstation from the set of the workstation working time sequence implicit coding feature vectors as the query feature vector, and use the other workstation working time sequence implicit coding feature vectors in the set of the workstation working time sequence implicit coding feature vectors as the library workstation working time sequence implicit coding feature vectors to obtain a set of library workstation working time sequence implicit coding feature vectors; a selection range ratio guided workstation working time feature invisible query response module is used to select the query feature vector and the set of the library workstation working time sequence implicit coding feature vectors to select the workstation working time feature invisible query response guided by the range ratio to obtain the workstation working time feature query response coding vector; a work time abnormality detection result judgment module is used to obtain a detection result indicating whether there is an abnormality in the working time of the first workstation based on the workstation working time feature query response coding vector.

10. The manufacturing execution management system for a production workshop according to claim 9, characterized in that: The workstation historical working time sequence encoding module is used to: use a sequence encoder to encode the historical working time data of each workstation to obtain a set of implicitly encoded feature vectors of the workstation working time sequence.

Images