Intelligent factory management method and management system

Abstract

The application relates to the field of smart factory management, and discloses a smart factory management method and a management system, which comprises the following steps: synchronously collecting the operation behavior, operation strength, task completion time and working environment state of personnel at each production post in a continuous production cycle to form a multi-dimensional time sequence feature set; performing nonlinear correlation analysis to construct a load response mapping model; performing multi-dimensional weighted comparison on the running states of different posts and personnel in the same production cycle to generate a key post and key personnel set; combining the expansion trend of the collaborative deviation section on the time axis, the load sensitivity reversal interval, the post bearing margin change and the historical operation stability of each personnel to generate a dynamic configuration adjustment scheme; and performing constraint regulation on the personnel post distribution and operation participation sequence in the production process to form a dynamic state update result. The application has the advantages of improving efficiency and stability.

Description

Technical Field

[0001] This invention relates to the field of smart factory management, specifically to a smart factory management method and system. Background Technology

[0002] In the field of smart factory production management, existing production scheduling technologies, during periods of increased workload, typically employ task allocation strategies aimed at maximizing task consumption based on the historical processing capacity or average output data of each workstation. The underlying concept is to improve overall production efficiency by fully utilizing existing capacity. However, these technologies neglect the changing characteristics of individual production positions under different times and workloads, failing to reflect the impact of work intensity and load fluctuations on the operational status of workstations during continuous operation. This can easily lead to task backlogs at individual workstations under high load conditions, causing imbalances in coordination between upstream and downstream workstations, disrupting the overall production rhythm, and reducing the operational efficiency and stability of the production system. Therefore, existing technologies struggle to accurately depict the dynamic evolution of production positions and to detect and regulate production rhythm disruptions caused by uneven workloads in a timely manner, thus failing to meet the management requirements of smart factories for high efficiency, stability, and controllability in the production process. Summary of the Invention

[0003] To address the shortcomings of existing technologies, this invention provides a smart factory management method and system, which has the advantages of improved efficiency and stability, and solves the problems mentioned in the background technology.

[0004] To achieve the aforementioned goals of improving efficiency and stability, this invention provides the following technical solution: a smart factory management method, comprising the following steps:

[0005] The work behavior, operational intensity, task completion time, and work environment status of personnel in each production position are collected synchronously during the continuous production cycle to form a multi-dimensional time series feature set that characterizes the stability of individual work and the evolution of job operation behavior.

[0006] Nonlinear correlation analysis was performed on the multidimensional time series feature set, and the response relationship between job task density, work continuity intensity, personal historical work stability and actual completion efficiency was fitted to construct a load response mapping model of job and personnel operation characteristics under different work pressures.

[0007] Based on the load response mapping model, the operating status of different positions and personnel within the same production cycle is compared in a multi-dimensional weighted manner, and a set of key positions and key personnel that have a substantial impact on the production rhythm is generated according to the load peak, task completion efficiency and collaborative deviation contribution rate.

[0008] Based on the set of key positions and key personnel, combined with the expansion trend of the coordination deviation segment on the time axis, the reversal interval of load sensitivity, the changes in the position's load margin, and the historical operational stability of each person, a dynamic configuration adjustment plan is generated for each position and person.

[0009] Based on the dynamic configuration adjustment scheme, the distribution of personnel positions and the order of work participation in the production process are constrained and controlled. The operation behavior of positions and the historical work status curve of each person are continuously updated during the production cycle, forming a dynamic status update result that represents the convergence of production rhythm, operation stability and individual load reliability.

[0010] Preferably, the process of forming a multidimensional time-series feature set representing the stability of individual work and the evolution of job performance is as follows:

[0011] Multi-dimensional sensing and data acquisition modules are deployed at each production station, including motion sensors, environmental monitoring units, and work task recorders;

[0012] The operation actions, force, and working time of each person in each position are collected synchronously at the microsecond level, and the environmental conditions are collected in combination with the temperature, humidity, light intensity and noise level around the position.

[0013] The collected raw data is processed by timestamp alignment, outlier removal and standardization, and the features of each dimension are aggregated according to time series to form a multi-dimensional time series feature set.

[0014] Preferably, the process of performing nonlinear correlation analysis on a multidimensional time-series feature set is as follows:

[0015] Using a multi-dimensional time-series feature set as input, the work behavior, operational intensity, task completion time, and environmental status data of each position and personnel are standardized.

[0016] Based on standardized data, a combination of multivariate nonlinear regression and deep neural network was used to fit the response relationship between job task density, work duration intensity, personal historical work stability and actual completion efficiency.

[0017] During the fitting process, time series convolutional networks are used to extract the evolution features of work behavior, and a job-person mapping matrix is ​​constructed based on the job-person correspondence extracted from the multi-dimensional time series feature set to encode the collaborative relationship between different jobs.

[0018] The output of the deep neural network is subjected to nonlinear optimization and iteration to generate prediction results of the operating status of positions and personnel under different work pressures.

[0019] Preferably, the process of constructing a load response mapping model for the operational characteristics of positions and personnel under different work pressures is as follows:

[0020] Using the operational status prediction results as input, the work behavior, operational intensity, task completion time and environmental status indicators of each position and personnel in the prediction results are normalized.

[0021] The normalized prediction results are mapped to a predefined load function domain to calculate the load value of each position and person under different work pressure conditions, and a position and person load matrix is ​​constructed.

[0022] Based on the job and personnel load matrix, and combined with the characteristics of the prediction results changing over time, the dynamic response relationship of job and personnel load under different work pressure conditions is modeled to form a load response mapping model.

[0023] Preferably, the process of performing a multi-dimensional weighted comparison of the operational status of different positions and personnel within the same production cycle is as follows:

[0024] Based on the load response mapping model, the load indicators of each position and person under different work pressure conditions are time-aligned with the work status data collected synchronously during the actual production cycle.

[0025] For the aligned data, calculate the load deviation, task completion efficiency deviation, and cross-position collaboration differences for each position and personnel;

[0026] The deviation, efficiency deviation and collaboration difference are weighted and integrated according to preset weights to form a multi-dimensional comparison matrix containing job ID, personnel ID and weighted values ​​of each dimension.

[0027] Based on the multidimensional comparison matrix, combined with the historical operational stability, operational intensity trend and cumulative task delay value of each position and personnel, dynamic correction is performed according to the preset weighting rules to generate a comprehensive indicator reflecting the deviation of the actual production operation status.

[0028] Preferably, the process of generating a set of key positions and key personnel that have a substantial impact on the production rhythm is as follows:

[0029] Using the multidimensional comparison results and the dynamically corrected comprehensive indicators as input, the peak load, task completion efficiency deviation and collaboration deviation contribution rate of each position and person are quantitatively calculated item by item to form a preliminary deviation score.

[0030] The initial deviation scores are weighted and merged according to preset weights to generate a comprehensive deviation score for each position and person, and a weighted sorted list is generated according to the position category and person number.

[0031] Based on the comprehensive deviation score and weighted ranking list, the comprehensive score is dynamically corrected by combining the historical work stability, operation intensity trend and task delay accumulation of each position and personnel, generating a corrected comprehensive score, and updating the weighted ranking order at the same time.

[0032] Based on the revised comprehensive score and the updated weighted ranking list, positions and personnel with high peak load, significant deviation in task completion efficiency, and large contribution rate of collaboration deviation are selected according to preset thresholds and marked as key positions and key personnel sets.

[0033] Preferably, the process of generating dynamic configuration adjustment plans for each position and personnel is as follows:

[0034] Based on the set of key positions and key personnel, the parameters of the preliminary control plan will be determined for each position and personnel according to the revised comprehensive score and weighted ranking.

[0035] The job allocation and work participation order in the preliminary control plan parameters are processed in segments, and the job positions and personnel with high deviation scores are given priority for adjustment operations.

[0036] During production execution, real-time data is collected on the work behavior, workload, task completion status, and collaboration information of each position and personnel.

[0037] The real-time collected data is compared and analyzed with the parameters of the preliminary control plan to calculate the load deviation and coordination anomaly value for each position and personnel.

[0038] For positions and personnel with high deviations and outliers, iterative adjustments are made, the order of position allocation and work participation is rearranged, and the control parameters are updated;

[0039] The updated job assignments and work sequences are validated and calculated to assess potential load peaks and coordination deviations. The validation results are then fed back and corrected with the updated comprehensive score to form the final dynamic configuration adjustment plan for each job and personnel.

[0040] Preferably, the process of constraining and controlling the distribution of personnel positions and the order of work participation in the production process is as follows:

[0041] The job assignment order and work participation order in the dynamic configuration adjustment plan will be sent to the production control system.

[0042] During the production process, monitor the distribution of personnel in each position in real time to detect vacancies, duplicate personnel allocation, or deviations in the work sequence.

[0043] By optimizing the scheduling algorithm to reallocate personnel to their posts and using the job sequence adjustment module to reorder the order of job participation, a revised job allocation and job sequence scheme is generated.

[0044] Preferably, the process of generating dynamic state update results that characterize production rhythm convergence, operational stability, and individual load reliability is as follows:

[0045] Based on the revised job allocation and work sequence plan, real-time data collection is conducted on the work behavior, operational intensity, task completion status, and collaboration information of each job and individual.

[0046] The collected data will be compared and analyzed with the correction plan and historical operation status curves;

[0047] Based on the comparative analysis results, the historical work status curves of positions and personnel are updated incrementally, and changes in indicators such as peak load, work efficiency, coordination deviation and operational intensity trends are recorded.

[0048] Based on the updated state curves, the production rhythm convergence, job operation stability, and personnel load reliability indicators are calculated to form dynamic state update results that characterize production rhythm convergence, operation stability, and individual load reliability.

[0049] A smart factory management system includes:

[0050] Behavior perception module: Synchronously collects the work behavior, operation intensity, task completion time and work environment status of personnel in various positions to form a multi-dimensional temporal feature set;

[0051] Load modeling module: Performs nonlinear correlation analysis on multidimensional time series features, fits the response relationship between job task density, work intensity and individual historical stability to actual completion efficiency, and constructs a load response mapping model;

[0052] Key identification module: Based on the load response model, it performs multi-dimensional weighted comparison of the operating status of positions and personnel to generate a set of key positions and key personnel that have the greatest impact on the production rhythm;

[0053] Solution generation module: Based on the set of key positions and personnel, combined with the trend of coordination deviation, load sensitivity range and position capacity margin, it generates a dynamic configuration adjustment plan for positions and personnel.

[0054] Status update module: It controls the distribution of positions and the order of operations according to the dynamic configuration scheme, and continuously updates the operation behavior of positions and the historical status curves of personnel to form dynamic status update results.

[0055] Compared with existing technologies, the present invention provides a smart factory management method and management system, which has the following beneficial effects:

[0056] This invention synchronously collects data on the work behavior, operational intensity, task completion time, and work environment of personnel in various production positions throughout a continuous production cycle. This forms a multi-dimensional time-series feature set characterizing individual work stability and the evolution of job performance. Nonlinear correlation analysis is used to fit the complex response relationship between job task density, work continuity intensity, individual historical work stability, and actual completion efficiency, constructing a load response mapping model of job and personnel operating characteristics under different work pressures. This enables quantitative analysis of personnel and job operating states. Based on this model, multi-dimensional weighted comparison identifies key positions and key personnel that substantially impact production rhythm. Furthermore, this is combined with the expansion trend of collaborative deviation segments on the time axis and load sensitivity... By analyzing sensitivity reversal intervals, changes in job load margins, and historical operational stability information, dynamic configuration adjustment schemes are generated for each job and individual. This allows for refined constraint and control of personnel job distribution and work participation sequence during the production process. Furthermore, the system continuously updates job operation behavior and individual work status curves throughout the production cycle, achieving dynamic optimization of production rhythm convergence, operational stability, and individual load reliability. By organically combining personnel work behavior characteristic collection, load response modeling, key job identification, and dynamic configuration control, a systematic technical approach is formed. Compared to traditional static scheduling and experience-based management, this approach offers significant technological advancements, including higher control precision, stronger adaptability, accurate identification of key jobs, and optimized production efficiency and personnel load management. Attached Figure Description

[0057] Figure 1 This is a schematic diagram of the method of the present invention;

[0058] Figure 2 This is a schematic diagram of the structure of the present invention. Detailed Implementation

[0059] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0060] Example 1: Please refer to Figure 1 As shown in the figure, a smart factory management method according to an embodiment of the present invention includes the following steps:

[0061] S1: Synchronously collect data on the work behavior, operational intensity, task completion time, and work environment status of personnel in each production position during a continuous production cycle to form a multi-dimensional temporal feature set that characterizes the stability of individual work and the evolution of job operation behavior.

[0062] The process of forming a multidimensional time-series feature set representing the evolution of individual job stability and job operation behavior in S1 is as follows:

[0063] Multi-dimensional sensing and data acquisition modules are deployed at each production station, including motion sensors, environmental monitoring units, and work task recorders;

[0064] High-precision motion sensors, such as accelerometers, gyroscopes, or pressure sensors, are installed at key positions in each job to collect personnel's operational movements and forces. At the same time, environmental monitoring units are deployed to collect environmental parameters such as temperature, humidity, light, and noise. A task recorder is used to record task type, execution sequence, and completion time. All acquisition modules are networked via industrial Ethernet or wireless industrial protocols to achieve real-time data transmission and unified management, ensuring that all types of data can be collected synchronously and meet microsecond-level time accuracy requirements.

[0065] The operation actions, force, and working time of each person in each position are collected synchronously at the microsecond level, and the environmental conditions are collected in combination with the temperature, humidity, light intensity and noise level around the position.

[0066] By using a unified timestamp to synchronously collect data output from each sensor, and through data buffering and clock synchronization mechanisms, it is ensured that each action data, operation force, and operation time are strictly aligned with the corresponding environmental parameters in the time dimension. At the same time, the start and end times of each person's operation are recorded in real time according to the task execution order and correlated with environmental monitoring data to provide accurate time reference for the generation of multi-dimensional features.

[0067] The collected raw data is processed by timestamp alignment, outlier removal and standardization, and the features of each dimension are aggregated according to time series to form a multi-dimensional time series feature set;

[0068] Data collected from different sensors are aligned using a unified timestamp to fill in any missing values. Then, statistical analysis methods, such as the three-standard-deviation method or median filtering, are used to remove outliers, ensuring data quality. Next, features of each dimension are standardized, including mean-variance normalization or minimum-maximum normalization, mapping data of different dimensions to a unified numerical range. This ensures the comparability and stability of multidimensional data in subsequent analysis and modeling. Furthermore, the actions, intensity, working time, and corresponding environmental conditions of each person at each position are organized into a matrix or tensor structure according to the collection time sequence. Rows represent time points, and columns represent different feature dimensions. By aggregating data through sliding windows or fixed time intervals, a high-dimensional time-series feature set reflecting the evolution of work behavior can be generated. This comprehensively characterizes individual work stability, work pattern evolution, and job behavior characteristics, providing a sufficient data foundation for nonlinear correlation analysis, load response modeling, and smart factory management decisions.

[0069] S2: Perform nonlinear correlation analysis on the multidimensional time series feature set, and fit the response relationship between job task density, work continuity intensity, personal historical work stability and actual completion efficiency to construct a load response mapping model of job and personnel operation characteristics under different work pressures.

[0070] The process of performing nonlinear correlation analysis on a multidimensional time series feature set in S2 is as follows:

[0071] Using a multi-dimensional time-series feature set as input, the work behavior, operational intensity, task completion time, and environmental status data of each position and personnel are standardized.

[0072] For each time series data point collected, missing values ​​are processed and outliers are detected. The mean-variance normalization or minimum-maximum normalization method is used to map the data of each dimension to a unified numerical range, thereby eliminating the influence of different feature units. After standardization, the data can maintain the comparability and stability of each feature in the subsequent modeling process, ensuring the convergence and prediction accuracy of nonlinear regression and deep neural network training.

[0073] Based on standardized data, a combination of multivariate nonlinear regression and deep neural network was used to fit the response relationship between job task density, work duration intensity, personal historical work stability and actual completion efficiency.

[0074] A multivariate nonlinear regression model is used to initially fit the historical data of various positions and personnel to obtain a rough functional relationship between task density and work efficiency. The output of the regression model is then jointly trained with the input features of a deep neural network. The deep neural network adopts a multi-layer fully connected structure or a residual structure, and the network parameters are iteratively optimized through the gradient descent algorithm. This enables the capture of complex nonlinear dependencies between multidimensional features, achieving accurate prediction of efficiency changes for individuals and positions under different work conditions.

[0075] During the fitting process, time series convolutional networks are used to extract the evolution features of work behavior, and a job-person mapping matrix is ​​constructed based on the job-person correspondence extracted from the multi-dimensional time series feature set to encode the collaborative relationship between different jobs.

[0076] By performing sliding window convolution on time series features through one-dimensional or two-dimensional convolution operations, the trends and patterns of job and personnel behavior over time can be extracted. The correspondence between job and personnel is encoded in matrix form, where the matrix elements represent the task allocation ratio, collaboration intensity, or historical collaborative performance between job and personnel. The behavioral evolution features output by the convolutional network are combined with the mapping matrix to form a high-dimensional collaborative feature vector that can be used for nonlinear optimization, thereby quantifying the mutual influence between job positions and the collaborative effect of personnel.

[0077] Nonlinear optimization and iteration are performed on the output of the deep neural network to generate prediction results of the operating status of positions and personnel under different work pressures.

[0078] Using the initial output of the deep neural network as the starting point for the predicted state, the predicted value is iteratively adjusted through nonlinear optimization algorithms, such as Newton's method, quasi-Newton method, or adaptive learning rate gradient optimization method, to minimize the residual between the predicted value and the historical observation data. During the iteration process, the weights of the optimization objective function can be dynamically adjusted according to the changes in job task density, work intensity, and personnel historical stability to enhance the robustness and adaptability of the prediction. Finally, the load response, task completion efficiency, and collaborative performance of each job and each person under different work pressures are output, providing a quantitative decision-making basis for the smart factory management system.

[0079] The process of constructing a load response mapping model for the operational characteristics of positions and personnel under different work pressures in S2 is as follows:

[0080] Using the operational status prediction results as input, the work behavior, operational intensity, task completion time and environmental status indicators of each position and personnel in the prediction results are normalized.

[0081] Data cleaning is performed on the job and personnel operation status prediction results obtained from deep neural networks or nonlinear regression. This includes removing outlier prediction values ​​and filling in any missing values. Standardization methods, such as mean-variance normalization or minimum-maximum normalization, are used to map various prediction indicators to a unified numerical range to eliminate the influence of differences in different feature units. This allows load calculations to be performed on a unified scale, improving calculation accuracy and comparability.

[0082] The normalized prediction results are mapped to a predefined load function domain to calculate the load value of each position and person under different work pressure conditions, and a position and person load matrix is ​​constructed.

[0083] Define the load function domain for positions and personnel, and input normalized work behavior indicators, operational intensity, task completion time, and environmental conditions into the load function. The function calculates the load value for each position and each person under different work pressure conditions, such as increased task density, increased operational intensity, or environmental deterioration. The load values ​​corresponding to each position and person are summarized to form a load matrix, where rows represent different positions or personnel and columns represent different work pressure conditions. The matrix structure quantifies the operational characteristics of positions and personnel under various pressure conditions.

[0084] Based on the job and personnel load matrix, and combined with the characteristics of the prediction results changing over time, the dynamic response relationship of job and personnel load under different work pressure conditions is modeled to form a load response mapping model.

[0085] By processing the load matrix using sliding time windows or time series analysis methods, the dynamic evolution characteristics of job and personnel loads as work pressure changes are captured. Combining the collaborative relationships between jobs and the historical work stability of personnel, multivariate nonlinear modeling methods, such as dynamic regression models, convolutional time series networks, or response mapping models based on deep neural networks, are used to fit the load response relationship. Finally, a load response mapping model is generated, which can predict the load change patterns of each job and personnel under different work pressure conditions, providing a quantitative decision-making basis for personnel allocation optimization, task scheduling, and load management in smart factories.

[0086] S3: Based on the load response mapping model, the operating status of different positions and personnel within the same production cycle is compared in a multi-dimensional weighted manner, and a set of key positions and key personnel that have a substantial impact on the production rhythm is generated according to the load peak, task completion efficiency and collaborative deviation contribution rate.

[0087] The process of performing a multi-dimensional weighted comparison of the operational status of different positions and personnel within the same production cycle in S3 is as follows:

[0088] Based on the load response mapping model, the load indicators of each position and person under different work pressure conditions are time-aligned with the work status data collected synchronously during the actual production cycle.

[0089] The system obtains the job and personnel load values ​​predicted by the load response mapping model, and extracts the actual work status data collected in real time by sensors and work recorders during the production cycle. The predicted load is strictly aligned with the actual work status through a unified timestamp to ensure that the load prediction at each time point matches the corresponding work action, operation intensity, task completion time and environmental status, providing an accurate time reference for deviation analysis.

[0090] For the aligned data, calculate the load deviation, task completion efficiency deviation, and cross-position collaboration differences for each position and personnel;

[0091] For each position and personnel, the load deviation is calculated at each time point, which is the difference or relative error between the actual load and the predicted load. At the same time, the task completion efficiency deviation is calculated and quantified by the difference between the actual task completion time and the standard working hours or expected completion time. In addition, the cross-position collaboration difference is calculated using the inter-position collaboration matrix and personnel collaboration relationship data to quantify the degree of deviation of task collaboration between different positions, so as to comprehensively reflect the abnormal operation status within the production cycle.

[0092] The deviation, efficiency deviation and collaboration difference are weighted and integrated according to preset weights to form a multi-dimensional comparison matrix containing job ID, personnel ID and weighted values ​​of each dimension.

[0093] Based on pre-set weights, the load deviation, task completion efficiency deviation, and collaboration difference are weighted and summed proportionally to generate a comprehensive index. The weighted value corresponding to each position and each person is associated with the position ID and the person ID to construct a multi-dimensional comparison matrix, where the rows represent positions and personnel, and the columns represent the weighted indicators of each dimension, thereby realizing a multi-dimensional quantitative comparison of the operating status of all positions and personnel within the same production cycle.

[0094] Based on the multidimensional comparison matrix, combined with the historical operational stability, operational intensity trend and cumulative task delay value of each position and personnel, dynamic correction is performed according to the preset weighting rules to generate a comprehensive indicator that reflects the deviation of the actual production operation status.

[0095] By comparing the current deviation with historical operational stability data, the deviation values ​​of positions or personnel with large fluctuations are appropriately smoothed. At the same time, the weights are dynamically adjusted by considering the trend of operational intensity and the cumulative value of task delays. This enables the comprehensive index to accurately reflect the abnormal load and efficiency deviation of positions and personnel in actual production. The final output comprehensive index can be used to identify high-risk positions, abnormal operation behaviors or potential collaboration bottlenecks, and provide a quantitative basis for smart factory scheduling optimization and management decisions.

[0096] The process of generating the set of key positions and key personnel that have a substantial impact on the production rhythm in S3 is as follows:

[0097] Using the multidimensional comparison results and the dynamically corrected comprehensive indicators as input, the peak load, task completion efficiency deviation and collaboration deviation contribution rate of each position and person are quantitatively calculated item by item to form a preliminary deviation score.

[0098] Key indicators for each position and personnel in the multidimensional comparison matrix are extracted, including peak load, task completion efficiency, and cross-position collaboration differences. Each indicator is normalized or standardized according to a unified quantitative standard to ensure that data from different dimensions can be directly compared. The degree of deviation of each position and personnel on each indicator is quantified through numerical calculation or simple summation to form a preliminary deviation score, which can intuitively reflect the magnitude of its potential impact on the production rhythm.

[0099] The initial deviation scores are weighted and merged according to preset weights to generate a comprehensive deviation score for each position and person, and a weighted sorted list is generated according to the position category and person number.

[0100] Based on enterprise management rules or production process requirements, weights are assigned to peak load, task completion efficiency deviation, and collaboration deviation contribution rate. For example, the peak load weight can be set to 0.4, the efficiency deviation weight to 0.35, and the collaboration deviation contribution rate weight to 0.25. The comprehensive deviation score for each position and personnel is calculated by weighted summation. The comprehensive scores are sorted according to position category, such as key production position, auxiliary position, and personnel number to form a weighted ranking list, providing an intuitive basis for identifying high-impact positions and personnel.

[0101] Based on the comprehensive deviation score and weighted ranking list, the comprehensive score is dynamically corrected by combining the historical work stability, operation intensity trend and task delay accumulation of each position and personnel, generating a corrected comprehensive score, and updating the weighted ranking order at the same time.

[0102] By using historical operational data, we analyze the stability fluctuations, operational intensity trends, and cumulative task delays of each position and personnel. We then dynamically adjust the preliminary comprehensive deviation score. For example, positions with stable historical performance and low task delays can have their deviation scores appropriately reduced, while positions with high volatility and high delays can have their scores appropriately increased. After adjustment, we regenerate the ranking list so that the ranking of key positions and personnel can more accurately reflect their potential impact on the production rhythm.

[0103] Based on the revised comprehensive score and the updated weighted ranking list, positions and personnel with high peak load, significant deviation in task completion efficiency, and large contribution rate of collaboration deviation are selected according to preset thresholds and marked as key positions and key personnel sets.

[0104] Set threshold standards, such as comprehensive scores above a certain percentile (e.g., the top 10%) or exceeding a specific score, and positions and personnel whose peak load and efficiency deviation exceed preset thresholds. By filtering the sorted list one by one, extract positions and personnel that meet the conditions, and mark these positions and personnel as key positions and key personnel sets for production scheduling optimization, risk warning and resource allocation decisions, so as to achieve quantitative management and control of production rhythm.

[0105] S4: Based on the set of key positions and key personnel, combined with the expansion trend of the coordination deviation segment on the time axis, the load sensitivity reversal interval, the changes in the position's load margin, and the historical operational stability of each person, generate a dynamic configuration adjustment plan for each position and person.

[0106] The process of generating dynamic configuration adjustment plans for each position and personnel in S4 is as follows:

[0107] Based on the set of key positions and key personnel, the parameters of the preliminary control plan will be determined for each position and personnel according to the revised comprehensive score and weighted ranking.

[0108] Extract the job ID and personnel ID from the set of key positions and key personnel, and combine them with their revised comprehensive scores and weighted rankings to assign preliminary control parameters to each position and personnel, including priority, work sequence, task allocation ratio, and work intensity control parameters. These parameters can be stored in matrix or table form for quick reference and dynamic adjustment during the control process.

[0109] The job allocation and work participation order in the preliminary control plan parameters are processed in segments, and the job positions and personnel with high deviation scores are given priority for adjustment operations.

[0110] The positions and personnel are ranked from highest to lowest according to the comprehensive deviation score. Positions or personnel with high deviation scores are divided into priority control segments. Within each control segment, the allocation of positions and the order of work participation are locally adjusted, such as arranging key operations in advance or redistributing task loads, to reduce potential load peaks and task delay risks. At the same time, the original order of positions and personnel with low deviation scores is retained, so as to achieve hierarchical optimization management.

[0111] During production execution, real-time data is collected on the work behavior, workload, task completion status, and collaboration information of each position and personnel.

[0112] By deploying motion sensors, work recorders, and collaborative monitoring units at various positions, the system collects each person's operational actions, force, task start and end times, as well as time differences, dependencies, and collaborative completion status during collaboration with other positions at the microsecond or millisecond level. The collected data is uploaded to the central processing unit in real time for dynamic comparison and analysis with the control plan.

[0113] The real-time collected data is compared and analyzed with the parameters of the preliminary control plan to calculate the load deviation and coordination anomaly value for each position and personnel.

[0114] By comparing real-time operation status data with the expected load and coordination targets in the preliminary control plan at each time point, load deviation (difference between actual load and planned load) and coordination anomalies (deviation of actual collaboration time or task dependency anomalies) are calculated. Through numerical and normalization processing, each indicator is mapped to a unified quantitative standard, providing a precise basis for iterative adjustments.

[0115] For positions and personnel with high deviations and outliers, iterative adjustments are made, the order of position allocation and work participation is rearranged, and the control parameters are updated;

[0116] Based on the magnitude of load deviation and coordination anomalies, the adjustment priority is determined, and the task sequence of positions or personnel with larger deviations is advanced or the workload is redistributed. During the adjustment process, iterative algorithms or rule engines are used to gradually optimize the allocation scheme. After each round of adjustment, the position allocation table, the work sequence table and related control parameters are updated to ensure that the dynamic configuration scheme can continuously respond to load fluctuations and coordination deviations that occur during the production process.

[0117] The updated job assignments and work sequences are validated and calculated to assess potential load peaks and coordination deviations. The validation results are then fed back and corrected with the updated comprehensive score to form the final dynamic configuration adjustment plan for each job and personnel.

[0118] The adjusted job assignments and work sequences are verified using simulation or prediction models. Potential load peaks, task delays, and collaboration anomalies are calculated. The comprehensive score is then further corrected, ultimately forming a complete dynamic configuration adjustment scheme that has undergone multiple rounds of verification and dynamic feedback. This scheme includes optimized work sequences, task allocation ratios, and load control parameters for each job and personnel. It can be directly applied to production execution to optimize production rhythm, balance load, and improve collaboration efficiency.

[0119] S5: Based on the dynamic configuration adjustment scheme, the distribution of personnel positions and the order of work participation in the production process are constrained and controlled. The operation behavior of positions and the historical work status curve of each person are continuously updated in the production cycle, forming a dynamic status update result that represents the convergence of production rhythm, operation stability and individual load reliability.

[0120] The process of constraining and controlling the distribution of personnel positions and the order of work participation in the production process in S5 is as follows:

[0121] The job assignment order and work participation order in the dynamic configuration adjustment plan will be sent to the production control system.

[0122] Based on the dynamic configuration adjustment plan generated for each position and personnel, the optimized position allocation order and work participation order are transmitted to the production control system or scheduling terminal in the form of electronic instructions or configuration files. This ensures that the control system can obtain the personnel arrangement and work execution order of each position in real time, providing a data foundation for constraint control.

[0123] During the production process, monitor the distribution of personnel in each position in real time to detect vacancies, duplicate personnel allocation, or deviations in the work sequence.

[0124] The production control system uses deployed job status sensing modules and work behavior sensing units to collect real-time information on personnel on duty, work execution status, and timestamps at each job position. It then compares and analyzes this information with the predetermined job allocation and work sequence in the dynamic configuration and adjustment plan. This allows the system to automatically identify abnormal situations such as job vacancies, duplicate personnel allocation, and deviations or delays in work sequence, providing a monitoring basis for dynamic correction.

[0125] By optimizing the scheduling algorithm to reallocate personnel to their posts and using the job sequence adjustment module to reorder the order of job participation, a revised job allocation and job sequence scheme is generated.

[0126] When abnormalities in job distribution or work sequence are detected, the optimization scheduling algorithm is invoked to dynamically adjust the allocation of personnel to their positions based on job load, personnel skill matching, and collaboration requirements. At the same time, the work sequence adjustment module is used to reorder the order of work participation. The generated correction plan is sent to the execution end through the control system to achieve real-time constraint and control of personnel job distribution and work sequence, ensuring the stability of production rhythm and job load.

[0127] The process of generating dynamic state update results in S5, representing production rhythm convergence, operational stability, and individual load reliability, is as follows:

[0128] Based on the revised job allocation and work sequence plan, real-time data collection is conducted on the work behavior, operational intensity, task completion status, and collaboration information of each job and individual.

[0129] Through the job status perception module and the work behavior perception unit, the real-time work actions, operating force, task completion time and collaborative operation of personnel in each position are continuously collected. The data is then marked with timestamps, job IDs and personnel IDs to form traceable multi-dimensional time-series data, providing an accurate basis for dynamic status analysis.

[0130] The collected data will be compared and analyzed with the correction plan and historical operation status curves;

[0131] The real-time collected job and personnel operation data are compared hourly with the corrected job allocation order and operation participation order. At the same time, combined with historical operation status curves, the deviations of job load, operation efficiency, coordination deviation and operation intensity are analyzed to identify anomalies and trend changes, providing a basis for dynamic status updates.

[0132] Based on the comparative analysis results, the historical work status curves of positions and personnel are updated incrementally, and changes in indicators such as peak load, work efficiency, coordination deviation and operational intensity trends are recorded.

[0133] Based on the comparative analysis results, the historical operation status curves are incrementally updated according to job position and personnel dimensions. The updated content includes indicators such as load peak records, operation completion efficiency, collaborative operation deviation, and operation intensity change trends. At the same time, time series matrices or tensors are generated to ensure that the data structure is complete and can be used for analysis and closed-loop control.

[0134] Based on the updated state curves, the production rhythm convergence, job operation stability, and personnel load reliability indicators are calculated to form dynamic state update results that characterize the production rhythm convergence, operation stability, and individual load reliability.

[0135] By utilizing the updated historical work status curves, and through the load response mapping model and multidimensional weighted analysis method, the production rhythm convergence, work behavior stability, and load reliability scores of each position and personnel are calculated. The results are then summarized to generate a dynamic status update matrix or tensor, which can be used for real-time monitoring, optimized scheduling, and anomaly warning. It also provides closed-loop feedback data for the next round of dynamic configuration adjustment, ensuring that the production process is efficient, stable, and controllable.

[0136] Example 2: Please refer to Figure 2 As shown, a smart factory management system includes:

[0137] Behavior perception module: Synchronously collects the work behavior, operation intensity, task completion time and work environment status of personnel in various positions to form a multi-dimensional temporal feature set;

[0138] Load modeling module: Performs nonlinear correlation analysis on multidimensional time series features, fits the response relationship between job task density, work intensity and individual historical stability to actual completion efficiency, and constructs a load response mapping model;

[0139] Key identification module: Based on the load response model, it performs multi-dimensional weighted comparison of the operating status of positions and personnel to generate a set of key positions and key personnel that have the greatest impact on the production rhythm;

[0140] Solution generation module: Based on the set of key positions and personnel, combined with the trend of coordination deviation, load sensitivity range and position capacity margin, it generates a dynamic configuration adjustment plan for positions and personnel.

[0141] Status update module: It controls the distribution of positions and the order of operations according to the dynamic configuration scheme, and continuously updates the operation behavior of positions and the historical status curves of personnel to form dynamic status update results.

[0142] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0143] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A smart factory management method, characterized by, Includes the following steps: The work behavior, operational intensity, task completion time, and work environment status of personnel in each production position are collected synchronously during the continuous production cycle to form a multi-dimensional time series feature set that characterizes the stability of individual work and the evolution of job operation behavior. Nonlinear correlation analysis was performed on the multidimensional time series feature set, and the response relationship between job task density, work continuity intensity, personal historical work stability and actual completion efficiency was fitted to construct a load response mapping model of job and personnel operation characteristics under different work pressures. Based on the load response mapping model, the operating status of different positions and personnel within the same production cycle is compared in a multi-dimensional weighted manner, and a set of key positions and key personnel that have a substantial impact on the production rhythm is generated according to the load peak, task completion efficiency and collaborative deviation contribution rate. Based on the set of key positions and key personnel, combined with the expansion trend of the coordination deviation segment on the time axis, the reversal interval of load sensitivity, the changes in the position's load margin, and the historical operational stability of each person, a dynamic configuration adjustment plan is generated for each position and person. Based on the dynamic configuration adjustment scheme, the distribution of personnel positions and the order of work participation in the production process are constrained and controlled. The operation behavior of positions and the historical work status curve of each person are continuously updated during the production cycle, forming a dynamic status update result that represents the convergence of production rhythm, operation stability and individual load reliability.

2. The method of claim 1, wherein, The process of forming a multidimensional time-series feature set that characterizes the stability of individual work and the evolution of job performance is as follows: Multi-dimensional sensing and data acquisition modules are deployed at each production station, including motion sensors, environmental monitoring units, and work task recorders; The operation actions, force, and working time of each person in each position are collected synchronously at the microsecond level, and the environmental conditions are collected in combination with the temperature, humidity, light intensity and noise level around the position. The collected raw data is processed by timestamp alignment, outlier removal and standardization, and the features of each dimension are aggregated according to time series to form a multi-dimensional time series feature set.

3. The method of claim 2, wherein, The process of performing nonlinear correlation analysis on a multidimensional time series feature set is as follows: Using a multi-dimensional time-series feature set as input, the work behavior, operational intensity, task completion time, and environmental status data of each position and personnel are standardized. Based on standardized data, a combination of multivariate nonlinear regression and deep neural network was used to fit the response relationship between job task density, work duration intensity, personal historical work stability and actual completion efficiency. During the fitting process, time series convolutional networks are used to extract the evolution features of work behavior, and a job-person mapping matrix is ​​constructed based on the job-person correspondence extracted from the multi-dimensional time series feature set to encode the collaborative relationship between different jobs. The output of the deep neural network is subjected to nonlinear optimization and iteration to generate prediction results of the operating status of positions and personnel under different work pressures.

4. The method of claim 3, wherein, The process of constructing a load response mapping model for the operational characteristics of positions and personnel under different work pressures is as follows: Using the operational status prediction results as input, the work behavior, operational intensity, task completion time and environmental status indicators of each position and personnel in the prediction results are normalized. The normalized prediction results are mapped to a predefined load function domain to calculate the load value of each position and person under different work pressure conditions, and a position and person load matrix is ​​constructed. Based on the job and personnel load matrix, and combined with the characteristics of the prediction results changing over time, the dynamic response relationship of job and personnel load under different work pressure conditions is modeled to form a load response mapping model.

5. The method of claim 4, wherein, The process of performing a multi-dimensional weighted comparison of the operational status of different positions and personnel within the same production cycle is as follows: Based on the load response mapping model, the load indicators of each position and person under different work pressure conditions are time-aligned with the work status data collected synchronously during the actual production cycle. For the aligned data, calculate the load deviation, task completion efficiency deviation, and cross-position collaboration differences for each position and personnel; The deviation, efficiency deviation and collaboration difference are weighted and integrated according to preset weights to form a multi-dimensional comparison matrix containing job ID, personnel ID and weighted values ​​of each dimension. Based on the multidimensional comparison matrix, combined with the historical operational stability, operational intensity trend and cumulative task delay value of each position and personnel, dynamic correction is performed according to the preset weighting rules to generate a comprehensive indicator reflecting the deviation of the actual production operation status.

6. The method of claim 5, wherein, The process of generating a set of key positions and key personnel that have a substantial impact on the production rhythm is as follows: Using the multidimensional comparison results and the dynamically corrected comprehensive indicators as input, the peak load, task completion efficiency deviation and collaboration deviation contribution rate of each position and person are quantitatively calculated item by item to form a preliminary deviation score. The initial deviation scores are weighted and merged according to preset weights to generate a comprehensive deviation score for each position and person, and a weighted sorted list is generated according to the position category and person number. Based on the comprehensive deviation score and weighted ranking list, the comprehensive score is dynamically corrected by combining the historical work stability, operation intensity trend and task delay accumulation of each position and personnel, generating a corrected comprehensive score, and updating the weighted ranking order at the same time. Based on the revised comprehensive score and the updated weighted ranking list, positions and personnel with high peak load, significant deviation in task completion efficiency, and large contribution rate of collaboration deviation are selected according to preset thresholds and marked as key positions and key personnel sets.

7. The method of claim 6, wherein, The process of generating dynamic configuration adjustment plans for each position and personnel is as follows: Based on the set of key positions and key personnel, the parameters of the preliminary control plan will be determined for each position and personnel according to the revised comprehensive score and weighted ranking. The job allocation and work participation order in the preliminary control plan parameters are processed in segments, and the positions and personnel with high deviation scores are given priority for adjustment operations. During production execution, real-time data is collected on the work behavior, workload, task completion status, and collaboration information of each position and personnel. The real-time collected data is compared and analyzed with the parameters of the preliminary control plan to calculate the load deviation and coordination anomaly value for each position and personnel. For positions and personnel with high deviations and outliers, iterative adjustments are made, the order of position allocation and work participation is rearranged, and the control parameters are updated; The updated job assignments and work sequences are validated and calculated to assess potential load peaks and coordination deviations. The validation results are then fed back and corrected with the updated comprehensive score to form the final dynamic configuration adjustment plan for each job and personnel.

8. The method of claim 7, wherein, The process of constraining and regulating the distribution of personnel in different positions and the order of their participation in the production process is as follows: The job assignment order and work participation order in the dynamic configuration adjustment plan will be sent to the production control system. During the production process, monitor the distribution of personnel in each position in real time to detect vacancies, duplicate personnel allocation, or deviations in the work sequence. By optimizing the scheduling algorithm to reallocate personnel to their posts and using the job sequence adjustment module to reorder the order of job participation, a revised job allocation and job sequence scheme is generated.

9. The method of claim 8, wherein, The process of generating dynamic state update results that characterize production rhythm convergence, operational stability, and individual load reliability is as follows: Based on the revised job allocation and work sequence plan, real-time data is collected on the work behavior, operational intensity, task completion status, and collaboration information of each job and personnel. The collected data will be compared and analyzed with the correction plan and historical operation status curves; Based on the comparative analysis results, the historical work status curves of positions and personnel are updated incrementally, and changes in indicators such as peak load, work efficiency, coordination deviation and operational intensity trends are recorded. Based on the updated state curves, the production rhythm convergence, job operation stability, and personnel load reliability indicators are calculated to form dynamic state update results that characterize production rhythm convergence, operation stability, and individual load reliability.

10. A smart factory management system applied to the method according to any one of claims 1-9, characterized in that, include: Behavior perception module: Synchronously collects the work behavior, operation intensity, task completion time and work environment status of personnel in various positions to form a multi-dimensional temporal feature set; Load modeling module: Performs nonlinear correlation analysis on multidimensional time series features, fits the response relationship between job task density, work intensity and individual historical stability to actual completion efficiency, and constructs a load response mapping model; Key identification module: Based on the load response model, it performs multi-dimensional weighted comparison of the operating status of positions and personnel to generate a set of key positions and key personnel that have the greatest impact on the production rhythm; Solution generation module: Based on the set of key positions and personnel, combined with the trend of coordination deviation, load sensitivity range and position capacity margin, it generates a dynamic configuration adjustment plan for positions and personnel. Status update module: It controls the distribution of positions and the order of operations according to the dynamic configuration scheme, and continuously updates the operation behavior of positions and the historical status curves of personnel to form dynamic status update results.

Images