Toothpick intelligent production plan optimization method based on big data

Abstract

The invention discloses a toothpick intelligent production plan optimization method based on big data, and belongs to the technical field of industrial intelligent production management, and the method comprises the following steps: S1, obtaining a unified data set based on raw material, equipment and order heterogeneous data sources; s2, adopting a prophet time sequence prediction model to obtain a capacity demand at a future moment, determining a multi-objective optimization function with capacity, cost and quality as objectives, and solving to obtain a toothpick production parameter scheme; s3, through dynamic weight calculation and priority ranking, a dynamic production scheduling plan adaptive to actual production is obtained, and implementation is carried out according to the dynamic production scheduling plan; and S4, in the production process, collecting deviation data of the production process in real time, and obtaining a production plan with improved robustness through deviation attribution analysis and parameter iterative correction. By adopting the method, accurate matching of production resources and intelligent management and control of the whole process are realized.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of industrial intelligent production management, and particularly relates to a toothpick intelligent production plan optimization method based on big data. BACKGROUND

[0002] Traditional toothpick production plans rely on manual experience and static data, and have problems such as low efficiency and resource waste, while through the Internet of Things, cloud computing and data analysis technology, real-time data collection, integration and mining are provided for the whole production process.

[0003] However, the application of the above technology to the traditional production plan has the phenomenon of data island, and the multi-source heterogeneous data is difficult to integrate, the market demand prediction is inaccurate due to the dependence on experience decision, the production scheduling and raw material procurement are disconnected to cause inventory accumulation or insufficient supply, the process parameter adjustment lacks data support to cause resource waste and quality fluctuation, and the quality problem root cause cannot be traced through data analysis, and it is difficult to quickly respond to market changes and environmental protection policy requirements. SUMMARY

[0004] The purpose of the present application is to provide a toothpick intelligent production plan optimization method based on big data to solve the above technical problems.

[0005] To achieve the above purpose, the present application provides a toothpick intelligent production plan optimization method based on big data, comprising the following steps: S1, based on the heterogeneous data sources of raw materials, equipment and orders, a unified data set covering the whole production process is obtained; S2, based on the equipment data of the unified data set of step S1, a prior time series prediction model involving equipment state is used to obtain the capacity demand at future time, and a multi-objective optimization function with capacity, cost and quality as targets is determined, and a time series correlation-equipment state perception adaptive reinforcement learning multi-objective optimization algorithm is used to solve, to obtain a toothpick production parameter scheme meeting the future capacity demand; S3, based on the order data of the unified data set of step S1, the order delivery urgency, profit contribution, customer importance and order specification complexity are taken as influence factors, dynamic weight calculation and priority sorting are performed to obtain a dynamic production scheduling plan suitable for actual production, and the dynamic production scheduling plan is implemented; S4, based on the dynamic production scheduling plan of step S3, production process deviation data is collected in real time during the production process, and through deviation attribution analysis and parameter iteration correction, a production plan with improved robustness is obtained.

[0006] Preferably, the raw material data in the unified dataset in step S1 includes: the collection of wood required for toothpick production and historical purchase records; the equipment data includes: the operating parameters and maintenance logs of the toothpick forming machine, cutting machine and packaging machine; and the order data includes: the product specifications, delivery time, order quantity and customer level of customer orders.

[0007] Preferably, the specific steps of step S2 are as follows: S21. Based on the unified dataset from step S1, obtain historical toothpick production capacity data, equipment operating status data, and external correlation factor data. S22. Based on the device operating status data from step S21, generate a device status mask using a status marking method to obtain mask data. Its formula is: ; in, For time nodes; S23. Use the historical production capacity data of toothpicks from step S21 as input, and mask data. As a factor for equipment shutdown, external correlation factor data is used as an auxiliary input to the Oracle time series prediction model for training, resulting in a trained Oracle time series prediction model. S24. Based on the pre-planned time-series prediction model trained in step S23, the model uses time-series prediction methods to predict production capacity data within a preset future time period, thereby obtaining the production capacity demand at future moments. ; S25. Construct a multi-objective optimization function based on capacity, cost, and quality, where the capacity demand at future moments is considered. The production capacity objective function is set by using the deviation minimization modeling method. The calculation formula is: ; in, for Real-time production parameters Corresponding actual production capacity; Based on the raw material costs, equipment energy costs, and labor costs of toothpick production, a cost objective function is set using cost accounting modeling methods. The calculation formula is: ; in, For raw material costs; For equipment energy consumption costs; For labor costs; This represents the minimum cost per unit of production capacity. Based on the quality indicators of toothpicks, such as dimensional tolerance, breaking strength, and surface smoothness, a quality objective function is set using a compliance rate modeling method. The calculation formula is: ; in, The total number of quality indicators, and ; for Time of the first The compliance values ​​of the quality indicators, and ; This represents the minimum rate of non-compliance with quality standards. S26. Based on the historical toothpick production capacity data, equipment operation status data, and external correlation factor data from step S21, the equipment status mask from step S22. And the future production capacity requirements of step S24 The standardized input dataset is obtained, where the formula for standardizing the production capacity data is: ; in, This is to meet the standardized production capacity requirements; This represents the minimum capacity demand within a pre-defined future timeframe. This represents the maximum capacity demand within a pre-defined future timeframe. Quantification of quality indicators: This involves quantifying the quality indicators from step S25. Quantization interval is ; Device mask calibration: The calibration formula is: ; Where 0 represents normal and 1 represents fault / maintenance; Obtain the standardized input dataset The formula is: ; in, Historical cost data; S27. Standardized input dataset based on step S26 By using dynamic weight modeling, the real-time weights for device mask awareness are obtained, including the calculation of entropy weights. The formula is: ; in, , For the first The first sample The values ​​of each target; The formula for calculating the mask correction factor is: ; Dynamic weight normalization, where the production capacity target weight The calculation formula is: ; Cost target weight The calculation formula is: ; Quality target weights The calculation formula is: ; in, , , Based on step entropy weights The entropy weights of the calculated capacity, cost, and quality targets; for The constant-time capacity fluctuation coefficient, and ; for Time-cost sensitivity coefficient, and , for Real-time fluctuations in raw material prices. This is the benchmark unit price for raw materials; for The time-quality correlation coefficient, and , , representing the size-strength correlation coefficient, , representing the coefficient of correlation between size and smoothness. , representing the strength-smoothness correlation coefficient; For the normalized denominator, and ; S28. Based on the weights in step S27 and the objective function in step S25, a multi-objective optimization function is constructed using the function integration method. The calculation formula is as follows: ; A constraint system is constructed to obtain an optimization model based on mask, timing, and quality constraints. The model formula is as follows: ; in, Capacity based on equipment status; To achieve the required quality standards 99%; Adjustment range of parameters Adjust the threshold; Physical constraints for parameters; These are the production parameters, namely, feed rate, equipment speed, and drying temperature; S29, The constrained multi-objective optimization model constructed based on step S28, and the device state mask from step S22. and future production capacity requirements of step S24 Construct a reinforcement learning environment model, in which the state space The formula is: ; in, The parameters were generated for the previous moment; Three quality indicators; Action space The formula is: ; reward function The calculation formula is: ; in, This is the time-series smoothing penalty coefficient; value function The formula for maximizing time-series returns is: ; in, for Always in the zone Next, select an action. Then, the total cumulative discount rewards that can be obtained in the future; Value judgments prior to the action update; The value after the action; Discount factor; An iterative process of adaptive exploration, exploration probability The calculation formula is: ; in, for The probability of exploration at any given moment; The initial exploration probability; It is a natural exponential function; This represents the cumulative number of iterations. The maximum total number of iterations is preset; In each iteration, For new parameters of random exploration, choose Find the action with the highest value, verify the constraints, calculate the reward, and update. Value; preset the number of iterations, and filter for non-dominated solutions to obtain the Pareto front solution set, the formula is: ; Calculate the time-series abrupt change index for each solution in the Pareto front solution set. The calculation formula is: ; reserve ,in, The maximum allowed mutation threshold; With dynamic weighted average , , Assuming the target weights, calculate the distance to the positive ideal solution for each solution. Distance to the negative ideal solution And through proximity Sort to obtain The solution at the maximum value is then used to obtain a toothpick production parameter scheme that meets future production capacity requirements, optimizes costs, and achieves quality standards.

[0008] Preferably, the specific steps of step S3 are as follows: S31. Based on the unified dataset generated in step S1, the order data is cleaned and standardized to obtain a standardized order dataset. S32. For the four core influencing factors of order delivery urgency, profit contribution, customer importance and order specification complexity, a standardized quantification method is used to convert data of different dimensions into quantitative values ​​and input them into a standardized order dataset to obtain the four factor quantitative value matrix for each order. S33. Based on production targets and corporate strategy, determine the relative importance of the four core influencing factors to obtain the weight vector for each core influencing factor. S34. Based on the basic weight vector and real-time production data, a processing logic for dynamically adjusting coefficients to correct the basic weights is designed to obtain the dynamic weight vector corresponding to each core influencing factor. S35. Based on the four factor quantization value matrices from step S32 and the dynamic weight vector from step S34, through... The calculation method is used to obtain the order priority ranking results; S36. Based on the order priority ranking result of step S35 and the real-time production resource data of step S34, a dynamic production scheduling plan that adapts to the actual production is obtained by allocating production resources according to priority order.

[0009] Preferably, the real-time production data in step S34 includes production load rate, emergency order ratio, and equipment capacity utilization rate.

[0010] Preferably, the dynamic production schedule obtained in step S36 that adapts to the actual production includes: the production start time, end time, and equipment used for each order.

[0011] Preferably, step S3 also includes step S37, which is: while implementing the dynamic production schedule, the production status is monitored in real time. If an abnormal situation occurs, the process returns to step S34 to recalculate the dynamic weight and order priority, and to formulate a new dynamic production schedule.

[0012] Preferably, the specific steps of step S4 are as follows: S41. Based on the dynamic production scheduling and real-time monitoring data generated in step 36, collect deviation data of capacity, cost and quality by comparing actual values ​​with planned values, and form a real-time deviation dataset. S42. Based on the real-time deviation dataset in step S42 and the historical correlation data in the unified dataset in step 1, the core causes of the deviation and the corresponding deviation contribution are determined through multi-factor correlation analysis and root cause identification algorithm. Based on the deviation contribution, the influence intensity of the corresponding core cause on the deviation is judged to obtain the deviation attribution result. S43. Based on the deviation attribution results of step S42 and the current production parameters determined in step S29, adjust the equipment parameters according to the deviation factors, and satisfy the following conditions: and constraint; S44. Based on the corrected production parameters in step S43 and the predictive time series prediction model in step S2, predict the capacity, cost, and quality under the corrected parameters, and verify whether the corrected solution meets the robustness standard. S45. Based on the robustness standard set in step S44, if the standard is met, the corrected parameters and the adjusted production schedule are integrated to generate a robust production plan for subsequent production execution; otherwise, return to step S42 to perform deviation attribution analysis again, recheck whether key influencing factors are missing or attribution errors are made, analyze the reasons again and correct the parameters until the robustness standard is met, ensuring that the final plan can stably cope with production fluctuations.

[0013] Preferably, the robustness criteria for step S44 include the following conditions: Capacity standard: The absolute value of the deviation between actual capacity and planned capacity ≤ the preset allowable deviation; Cost standard: The absolute value of the deviation between the actual unit cost and the planned cost is less than or equal to the preset allowable deviation; Quality standard: Actual quality compliance rate ≥ Planned quality compliance rate.

[0014] Therefore, the above-mentioned method for optimizing intelligent toothpick production planning based on big data has the following beneficial effects: 1. By integrating three heterogeneous data sources—raw materials, equipment, and orders—a unified dataset covering the entire toothpick production process is formed. This effectively solves the problems of data fragmentation and information silos in traditional production. It also provides comprehensive, coherent, and standardized basic data support, avoiding information gaps in decision-making caused by scattered data sources and inconsistent formats, and ensuring that all subsequent optimization steps have reliable data basis.

[0015] 2. By introducing a predictive time-series model involving equipment status and combining it with equipment status masks, the accuracy of future capacity demand forecasting is improved, avoiding capacity prediction deviations caused by ignoring equipment downtime factors. At the same time, a multi-objective optimization function is constructed around capacity, cost, and quality, and solved through an adaptive reinforcement learning algorithm based on time-series correlation and equipment status awareness. This minimizes the deviation between actual and target capacity, reduces the comprehensive cost of raw materials, energy consumption, and labor per unit of capacity, and minimizes the rate of non-compliance with quality standards such as toothpick size tolerance, fracture strength, and surface smoothness. The final output production parameter scheme can accurately match future capacity demand, balancing optimal cost and stable quality.

[0016] 3. Using order delivery urgency, profit contribution, customer importance, and order specification complexity as core influencing factors, dynamic weight calculation and priority ranking break the rigid limitations of traditional fixed-weight production scheduling. The generated dynamic production schedule clearly defines the start and end times of production for each order and the equipment used, and can respond to abnormal situations through real-time production status monitoring. When an abnormality occurs, the weight and priority can be recalculated to adjust the production schedule. This ensures the priority delivery of urgent orders, high-profit orders, and orders from important customers, while also taking into account the production adaptability of complex specification orders, effectively improving order delivery efficiency and customer satisfaction, and reducing the waste of production capacity resources.

[0017] 4. By collecting deviation data on production capacity, cost, and quality in real time during the production process, and combining it with historical correlation data to conduct multi-factor correlation analysis and root cause identification, the core reasons for deviations can be accurately located. At the same time, the robustness of the correction scheme is verified through the Predictive Time Series Prediction Model, and iterative optimization is carried out until the standard is met. The final production plan can stably cope with various fluctuations in raw materials, equipment, orders, and other links, significantly improving the stability and risk resistance of the toothpick production process and reducing production losses caused by deviations.

[0018] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0019] Figure 1 The flowchart illustrates a method for optimizing intelligent toothpick production planning based on big data, as provided in this invention. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the embodiments of the present invention will be further described in detail below with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are merely illustrative of the embodiments of the present invention and are not intended to limit the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of this application. Examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout.

[0021] It should be noted that the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion, such as a process, method, system, product, or server that includes a series of steps or units, not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such process, method, product, or device.

[0022] The embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0023] The existing technological background is that traditional toothpick production planning relies heavily on manual experience or simple data processing methods, without fully integrating heterogeneous data sources and lacking systematic and intelligent consideration of the entire production process. This results in fragmented data, without forming a unified dataset covering the entire production process, leading to incomplete decision-making basis, and making it impossible to trace the root cause of quality problems through data analysis, making it difficult to respond quickly to market changes and environmental policy requirements.

[0024] Based on the above analysis, this invention is designed. (See appendix.) Figure 1 A method for optimizing intelligent toothpick production planning based on big data includes the following steps: S1. Based on heterogeneous data sources of raw materials, equipment, and orders, a unified dataset covering the entire production process is obtained; The raw material data in the unified dataset in step S1 includes: collecting the wood required for toothpick production and historical purchase records; equipment data includes: operating parameters and maintenance logs of toothpick forming machines, cutting machines, and packaging machines; and order data includes: product specifications, delivery deadlines, order quantities, and customer levels for customer orders.

[0025] S2. Based on the equipment data of the unified dataset in step S1, a predictive time-series prediction model involving equipment status is adopted to obtain the production capacity demand at future moments. At the same time, a multi-objective optimization function with production capacity, cost and quality as objectives is determined, and the toothpick production parameter scheme that meets the future production capacity demand is obtained by solving the adaptive reinforcement learning multi-objective optimization algorithm of time-series correlation and equipment status awareness. The specific steps of step S2 are as follows: S21. Based on the unified dataset from step S1, obtain historical toothpick production capacity data, equipment operating status data, and external correlation factor data. S22. Based on the device operating status data from step S21, generate a device status mask using a status marking method to obtain mask data. Its formula is: ; in, For time nodes; S23. Use the historical production capacity data of toothpicks from step S21 as input, and mask data. As a factor for equipment shutdown, external correlation factor data is used as an auxiliary input to the Oracle time series prediction model for training, resulting in a trained Oracle time series prediction model. S24. Based on the pre-planned time-series prediction model trained in step S23, the model uses time-series prediction methods to predict production capacity data within a preset future time period, thereby obtaining the production capacity demand at future moments. ; S25. Construct a multi-objective optimization function based on capacity, cost, and quality, where the capacity demand at future moments is considered. The production capacity objective function is set by using the deviation minimization modeling method. The calculation formula is: ; in, for Real-time production parameters Corresponding actual production capacity; Based on the raw material costs, equipment energy costs, and labor costs of toothpick production, a cost objective function is set using cost accounting modeling methods. The calculation formula is: ; in, For raw material costs; For equipment energy consumption costs; For labor costs; This represents the minimum cost per unit of production capacity. Based on the quality indicators of toothpicks, such as dimensional tolerance, breaking strength, and surface smoothness, a quality objective function is set using a compliance rate modeling method. The calculation formula is: ; in, The total number of quality indicators, and ; for Time of the first The compliance values ​​of the quality indicators, and ; This represents the minimum rate of non-compliance with quality standards. S26. Based on the historical toothpick production capacity data, equipment operation status data, and external correlation factor data from step S21, the equipment status mask from step S22. And the future production capacity requirements of step S24 The standardized input dataset is obtained, where the formula for standardizing the production capacity data is: ; in, This is to meet the standardized production capacity requirements; This represents the minimum capacity demand within a pre-defined future timeframe. This represents the maximum capacity demand within a pre-defined future timeframe. Quantification of quality indicators: This involves quantifying the quality indicators from step S25. Quantization interval is ; Device mask calibration: The calibration formula is: ; Where 0 represents normal and 1 represents fault / maintenance; Obtain the standardized input dataset The formula is: ; in, Historical cost data; S27. Standardized input dataset based on step S26 By using dynamic weight modeling, the real-time weights for device mask awareness are obtained, including the calculation of entropy weights. The formula is: ; in, , For the first The first sample The values ​​of each target; The formula for calculating the mask correction factor is: ; Dynamic weight normalization, where the production capacity target weight The calculation formula is: ; Cost target weight The calculation formula is: ; Quality target weights The calculation formula is: ; in, , , Based on step entropy weights The entropy weights of the calculated capacity, cost, and quality targets; for The constant-time capacity fluctuation coefficient, and ; for Time-cost sensitivity coefficient, and , for Real-time fluctuations in raw material prices. This is the benchmark unit price for raw materials; for The time-quality correlation coefficient, and , , representing the size-strength correlation coefficient, , representing the coefficient of correlation between size and smoothness. , representing the strength-smoothness correlation coefficient; For the normalized denominator, and ; S28. Based on the weights in step S27 and the objective function in step S25, a multi-objective optimization function is constructed using the function integration method. The calculation formula is as follows: ; A constraint system is constructed to obtain an optimization model based on mask, timing, and quality constraints. The model formula is as follows: ; in, Capacity based on equipment status; To achieve the required quality standards 99%; Adjustment range of parameters Adjust the threshold; Physical constraints for parameters; These are the production parameters, namely, feed rate, equipment speed, and drying temperature; S29, The constrained multi-objective optimization model constructed based on step S28, and the device state mask from step S22. and future production capacity requirements of step S24 Construct a reinforcement learning environment model, in which the state space The formula is: ; in, The parameters were generated for the previous moment; Three quality indicators; Action space The formula is: ; reward function The calculation formula is: ; in, This is the time-series smoothing penalty coefficient; value function The formula for maximizing time-series returns is: ; in, for Always in the zone Next, select an action. Then, the total cumulative discount rewards that can be obtained in the future; Value judgments prior to the action update; The value after the action; Discount factor; An iterative process of adaptive exploration, exploration probability The calculation formula is: ; in, for The probability of exploration at any given moment; The initial exploration probability; It is a natural exponential function; This represents the cumulative number of iterations. The maximum total number of iterations is preset; In each iteration, For new parameters of random exploration, choose Find the action with the highest value, verify the constraints, calculate the reward, and update. Value; preset the number of iterations, and filter for non-dominated solutions to obtain the Pareto front solution set, the formula is: ; Calculate the time-series abrupt change index for each solution in the Pareto front solution set. The calculation formula is: ; reserve ,in, The maximum allowed mutation threshold; With dynamic weighted average , , Assuming the target weights, calculate the distance to the positive ideal solution for each solution. Distance to the negative ideal solution And through proximity Sort to obtain The solution at the maximum value is then used to obtain a toothpick production parameter scheme that meets future production capacity requirements, optimizes costs, and achieves quality standards.

[0026] S3. Based on the order data of the unified dataset in step S1, and taking the urgency of order delivery, profit contribution, customer importance and order specification complexity as influencing factors, a dynamic production schedule is obtained through dynamic weight calculation and priority ranking, and then implemented according to the dynamic production schedule. The specific steps of step S3 are as follows: S31. Based on the unified dataset generated in step S1, the order data is cleaned and standardized, including removing invalid orders, adding key fields, and converting unstructured data to structured data to obtain a standardized order dataset. ; S32. For the four core influencing factors of order delivery urgency, profit contribution, customer importance, and order specification complexity, a standardized quantification method is used to convert data of different dimensions into quantified values, which are then input into a standardized order dataset to obtain a matrix of quantified values ​​for the four factors for each order; among which, order delivery urgency... The calculation formula is: ; in, This is the industry standard delivery cycle; For the first The remaining delivery time for each order; Profit contribution The calculation formula is: ; in, For the first Total profit of each order; For the first The production capacity requirement for each order; This represents the maximum unit capacity profit for all valid orders. Customer importance The calculation formula is: ; in, For the first The total amount of customer cooperation corresponding to each order; For the first The number of years of cooperation with the customer corresponding to each order; This represents the maximum total customer cooperation amount corresponding to all valid orders. This represents the maximum number of years of customer cooperation corresponding to all valid orders. Order specification complexity The calculation formula is: ; in, For the first The number of specification items per order; For the first The degree of customization of each order; The maximum value of the comprehensive complexity quantification among all orders; The order factor quantization matrix is ​​obtained using the following formula: ; S33. Based on production targets and corporate strategy, determine the relative importance of the four core influencing factors to obtain the weight vector for each core influencing factor: Based on the company's production goals, the basic weights of four factors are determined, and the sum of the weights is equal to 1, resulting in: ; S34. Based on the basic weight vector and real-time production data, a processing logic for dynamically adjusting coefficients to correct the basic weights is designed to obtain the dynamic weight vector corresponding to each core influence factor. The calculation formula is as follows: ; ; in, for Time of the first Dynamic weights of each factor; For the first State adjustment coefficients for each factor; for Real-time production status indicators at any given moment; Then dynamic weight These correspond to order delivery urgency, profit contribution, customer importance, and order specification complexity, respectively. The real-time production data in step S34 includes production load rate, proportion of urgent orders, and equipment capacity utilization rate; S35. Based on the four factor quantization value matrices from step S32 and the dynamic weight vector from step S34, through... The calculation method yields the order priority ranking result; the calculation formula is: ; in, For the first The priority score of each order; The order factor matrix is ​​the first... The first order One factor value; Arranged from highest to lowest, the order priority sequence is obtained: ; S36. Based on the order priority ranking result of step S35 and the real-time production resource data of step S34, a dynamic production scheduling plan adapted to the actual production situation is obtained by allocating production resources according to priority order; wherein, the formula for calculating the order processing time is: ; in, For the first One order in equipment Processing time; For the first The demand for each order; For equipment Productivity per unit time; For equipment Operating efficiency; The formula for calculating production time allocation is: ; ; in, For the first Production start time for each order; For priority higher than the first The set of all orders for a given order; For the first Production completion time for each order; To output dynamic production scheduling plans, ; The dynamic production schedule obtained in step S36, adapted to the actual production situation, includes: the start time, end time, and equipment used for each order. S37. While implementing the dynamic production schedule, the production status is also monitored in real time. If an abnormal situation occurs, return to step S34 to recalculate the dynamic weight and order priority, and formulate a new dynamic production schedule.

[0027] The specific steps of step S4 are as follows: S41. Based on the dynamic production scheduling and real-time monitoring data generated in step 36, by comparing actual values ​​with planned values, collect deviation data of capacity, cost, and quality to form a real-time deviation dataset, wherein capacity deviation... The calculation formula is: ; in, for Real-time actual production capacity; for Plan production capacity at all times; Cost deviation The calculation formula is: ; in, for Real-time unit production cost; for Real-time planning of unit production capacity cost; Quality deviation The calculation formula is: ; in, for Actual quality compliance rate at any given time; for Time plan quality compliance rate; Obtain the final deviation dataset The formula is: ; in, This refers to the start time of the production scheduling plan in step S36; This represents the current data collection time. S42. Based on the real-time deviation dataset from step S42 and the historical correlation data in the unified dataset from step 1, multi-factor correlation analysis and root cause identification algorithms are used to determine the core causes of the deviation and their corresponding deviation contributions. Using the deviation contribution as a benchmark, the influence strength of the corresponding core causes on the deviation is judged to obtain the deviation attribution result. The contribution calculation formula is as follows: ; in, For the first The contribution of each potential influencing factor to the bias; the larger the value, the stronger the influence of that factor on the bias. For the first One potential influencing factor; based on The magnitude of the value is used to determine the core influencing factors and the causes of the deviation; S43. Based on the deviation attribution results of step S42 and the current production parameters determined in step S29, adjust the equipment parameters according to the deviation factors, while satisfying constraints g3 and g4; the parameter correction formula is: ; in, For the first Corrected values ​​for each production parameter; For the first The current value of each production parameter; This is a correction factor; For the first The first core influencing factor affects the... The correlation contribution of each production parameter; This is the current dominant deviation value; This represents the maximum value of similar deviations in history. S44. Based on the corrected production parameters in step S43 and the predictive time series forecasting model in step S2, predict the capacity, cost, and quality under the corrected parameters, and verify whether the corrected solution meets the robustness standard: The robustness criteria for step S44 include the following conditions: Capacity standard: The absolute value of the deviation between actual capacity and planned capacity ≤ the preset allowable deviation; Cost standard: The absolute value of the deviation between the actual unit cost and the planned cost is less than or equal to the preset allowable deviation; Quality standard: Actual quality compliance rate ≥ Planned quality compliance rate; S45. Based on the robustness standard set in step S44, if the standard is met, the corrected parameters and the adjusted production schedule are integrated to generate a robust production plan for subsequent production execution; otherwise, return to step S42 to perform deviation attribution analysis again, recheck whether key influencing factors are missing or attribution errors are made, analyze the reasons again and correct the parameters until the robustness standard is met, ensuring that the final plan can stably cope with production fluctuations.

[0028] In summary, the big data-based intelligent toothpick production planning optimization method specifically addresses the core problems in traditional toothpick production: "capacity forecasting deviates from actual equipment conditions, leading to underproduction / overproduction; production scheduling relies solely on manual experience, ignoring order value and production fluctuations; and production deviations are difficult to pinpoint root causes, resulting in weak planning robustness." It achieves accurate capacity forecasting through a predictive time-series model combined with equipment status masks. A multi-objective optimization algorithm determines production parameters that balance capacity, cost, and quality. The dynamic scheduling process uses multi-factor quantification of orders and real-time weight adjustments to intelligently adapt order priorities. Furthermore, multi-factor correlation analysis and root cause analysis accurately locate production deviations and iteratively correct parameters. Ultimately, this method ensures a high degree of alignment between production plans and capacity demands and order value, while also improving the robustness of the plan in dealing with equipment failures and order fluctuations. This effectively enhances the efficiency, cost controllability, and customer delivery satisfaction in toothpick production.

[0029] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the technical solutions of the present invention, and these modifications or equivalent substitutions cannot cause the modified technical solutions to deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for optimizing intelligent toothpick production planning based on big data, characterized in that: Includes the following steps: S1. Based on heterogeneous data sources of raw materials, equipment, and orders, a unified dataset covering the entire production process is obtained; S2. Based on the equipment data of the unified dataset in step S1, a predictive time-series prediction model involving equipment status is adopted to obtain the production capacity demand at future moments. At the same time, a multi-objective optimization function with production capacity, cost and quality as objectives is determined, and the toothpick production parameter scheme that meets the future production capacity demand is obtained by solving the adaptive reinforcement learning multi-objective optimization algorithm of time-series correlation and equipment status awareness. S3. Based on the order data of the unified dataset in step S1, and taking the urgency of order delivery, profit contribution, customer importance and order specification complexity as influencing factors, a dynamic production schedule is obtained through dynamic weight calculation and priority ranking, and then implemented according to the dynamic production schedule. S4. Based on the dynamic production scheduling plan in step S3, during the production process, deviation data of the production process is collected in real time, and a production plan with improved robustness is obtained through deviation attribution analysis and parameter iterative correction.

2. The method for optimizing intelligent toothpick production planning based on big data according to claim 1, characterized in that: The raw material data in the unified dataset in step S1 includes: collecting the wood required for toothpick production and historical purchase records; equipment data includes: operating parameters and maintenance logs of toothpick forming machines, cutting machines, and packaging machines; and order data includes: product specifications, delivery deadlines, order quantities, and customer levels for customer orders.

3. The method for optimizing intelligent toothpick production planning based on big data according to claim 2, characterized in that: The specific steps of step S2 are as follows: S21. Based on the unified dataset from step S1, obtain historical toothpick production capacity data, equipment operating status data, and external correlation factor data. S22. Based on the device operating status data from step S21, generate a device status mask using a status marking method to obtain mask data. Its formula is: ； in, For time nodes; S23. Use the historical production capacity data of toothpicks from step S21 as input, and mask data. As a factor for equipment shutdown, external correlation factor data is used as an auxiliary input to the Oracle time series prediction model for training, resulting in a trained Oracle time series prediction model. S24. Based on the pre-planned time-series prediction model trained in step S23, the model uses time-series prediction methods to predict production capacity data within a preset future time period, thereby obtaining the production capacity demand at future moments. ; S25. Construct a multi-objective optimization function based on capacity, cost, and quality, where the capacity demand at future moments is considered. The production capacity objective function is set by using the deviation minimization modeling method. The calculation formula is: ； in, for Real-time production parameters Corresponding actual production capacity; Based on the raw material costs, equipment energy costs, and labor costs of toothpick production, a cost objective function is set using cost accounting modeling methods. The calculation formula is: ； in, For raw material costs; For equipment energy consumption costs; For labor costs; This represents the minimum cost per unit of production capacity. Based on the quality indicators of toothpicks, such as dimensional tolerance, breaking strength, and surface smoothness, a quality objective function is set using a compliance rate modeling method. The calculation formula is: ； in, The total number of quality indicators, and ; for Time of the first The compliance values ​​of the quality indicators, and ; This represents the minimum rate of non-compliance with quality standards. S26. Based on the historical toothpick production capacity data, equipment operation status data, and external correlation factor data from step S21, the equipment status mask from step S22. And the future production capacity requirements of step S24 The standardized input dataset is obtained, where the formula for standardizing the production capacity data is: ； in, This is to meet the standardized production capacity requirements; This represents the minimum capacity demand within a pre-defined future timeframe. This represents the maximum capacity demand within a pre-defined future timeframe. Quantification of quality indicators: This involves quantifying the quality indicators from step S25. Quantization interval is ; Device mask calibration: The calibration formula is: ; Where 0 represents normal and 1 represents fault / maintenance; Obtain the standardized input dataset The formula is: ； in, Historical cost data; S27. Standardized input dataset based on step S26 By using dynamic weight modeling, the real-time weights for device mask awareness are obtained, including the calculation of entropy weights. The formula is: ； in, , For the first The first sample The values ​​of each target; The formula for calculating the mask correction factor is: ； Dynamic weight normalization, where the production capacity target weight The calculation formula is: ； Cost target weight The calculation formula is: ； Quality target weights The calculation formula is: ； in, , , Based on step entropy weights The entropy weights of the calculated capacity, cost, and quality targets; for The constant-time capacity fluctuation coefficient, and ; for Time-cost sensitivity coefficient, and , for Real-time fluctuations in raw material prices. The base unit price of raw materials; for The time-quality correlation coefficient, and , , representing the size-strength correlation coefficient, , representing the coefficient of correlation between size and smoothness. , representing the strength-smoothness correlation coefficient; For the normalized denominator, and ; S28. Based on the weights in step S27 and the objective function in step S25, a multi-objective optimization function is constructed using the function integration method. The calculation formula is as follows: ； A constraint system is constructed to obtain an optimization model based on mask, timing, and quality constraints. The model formula is as follows: ； in, Capacity based on equipment status; To achieve the required quality standards 99%; Adjustment range of parameters Adjust the threshold; Physical constraints for parameters; These are the production parameters, namely, feed rate, equipment speed, and drying temperature; S29, The constrained multi-objective optimization model constructed based on step S28, and the device state mask from step S22. and future production capacity requirements of step S24 Construct a reinforcement learning environment model, in which the state space The formula is: ； in, The parameters were generated for the previous moment; Three quality indicators; Action space The formula is: ； reward function The calculation formula is: ； in, This is the time-series smoothing penalty coefficient; value function The formula for maximizing time-series returns is: ； in, for Always in the zone Next, select an action. Then, the total cumulative discount rewards that can be obtained in the future; Value judgments prior to the action update; The value after the action; Discount factor; An iterative process of adaptive exploration, exploration probability The calculation formula is: ； in, for The probability of exploration at any given moment; This represents the initial exploration probability; It is a natural exponential function; This represents the cumulative number of iterations. The maximum total number of iterations is preset; In each iteration, For new parameters of random exploration, choose Find the action with the highest value, verify the constraints, calculate the reward, and update. Value; preset the number of iterations, and filter for non-dominated solutions to obtain the Pareto front solution set, the formula is: ； Calculate the time-series abrupt change index for each solution in the Pareto front solution set. The calculation formula is: ； reserve ,in, The maximum allowed mutation threshold; With dynamic weighted average , , Assuming the target weights, calculate the distance to the positive ideal solution for each solution. Distance to the negative ideal solution And through proximity Sort to obtain The solution at the maximum value is then used to obtain a toothpick production parameter scheme that meets future production capacity requirements, optimizes costs, and achieves quality standards.

4. The method for optimizing intelligent toothpick production planning based on big data according to claim 3, characterized in that: The specific steps of step S3 are as follows: S31. Based on the unified dataset generated in step S1, the order data is cleaned and standardized to obtain a standardized order dataset. S32. For the four core influencing factors of order delivery urgency, profit contribution, customer importance and order specification complexity, a standardized quantification method is used to convert data of different dimensions into quantitative values ​​and input them into a standardized order dataset to obtain the four factor quantitative value matrix for each order. S33. Based on production targets and corporate strategy, determine the relative importance of the four core influencing factors to obtain the weight vector for each core influencing factor. S34. Based on the basic weight vector and real-time production data, a processing logic for dynamically adjusting coefficients to correct the basic weights is designed to obtain the dynamic weight vector corresponding to each core influencing factor. S35. Based on the four factor quantization value matrices from step S32 and the dynamic weight vector from step S34, through... The calculation method is used to obtain the order priority ranking results; S36. Based on the order priority ranking result of step S35 and the real-time production resource data of step S34, a dynamic production scheduling plan that adapts to the actual production is obtained by allocating production resources according to priority order.

5. The method for optimizing intelligent toothpick production planning based on big data according to claim 4, characterized in that: The real-time production data in step S34 includes production load rate, emergency order percentage, and equipment capacity utilization rate.

6. The method for optimizing intelligent toothpick production planning based on big data according to claim 5, characterized in that: The dynamic production schedule obtained in step S36, which is adapted to the actual production situation, includes: the start time, end time, and equipment used for each order.

7. The method for optimizing intelligent toothpick production planning based on big data according to claim 6, characterized in that: Step S3 also includes step S37, which is: while implementing the dynamic production schedule, the production status is monitored in real time. If an abnormal situation occurs, the process returns to step S34 to recalculate the dynamic weight and order priority, and to formulate a new dynamic production schedule.

8. The method for optimizing intelligent toothpick production planning based on big data according to claim 7, characterized in that: The specific steps of step S4 are as follows: S41. Based on the dynamic production scheduling and real-time monitoring data generated in step 36, collect deviation data of capacity, cost and quality by comparing actual values ​​with planned values, and form a real-time deviation dataset. S42. Based on the real-time deviation dataset in step S42 and the historical correlation data in the unified dataset in step 1, the core causes of the deviation and the corresponding deviation contribution are determined through multi-factor correlation analysis and root cause identification algorithm. Based on the deviation contribution, the influence intensity of the corresponding core cause on the deviation is judged to obtain the deviation attribution result. S43. Based on the deviation attribution results of step S42 and the current production parameters determined in step S29, adjust the equipment parameters according to the deviation factors, and satisfy the following conditions: and constraint; S44. Based on the corrected production parameters in step S43 and the predictive time series prediction model in step S2, predict the capacity, cost, and quality under the corrected parameters, and verify whether the corrected solution meets the robustness standard. S45. Based on the robustness standard set in step S44, if the standard is met, the corrected parameters and the adjusted production schedule are integrated to generate a robust production plan for subsequent production execution; otherwise, return to step S42 to perform deviation attribution analysis again, recheck whether key influencing factors are missing or attribution errors are made, analyze the reasons again and correct the parameters until the robustness standard is met, ensuring that the final plan can stably cope with production fluctuations.

9. The method for optimizing intelligent toothpick production planning based on big data according to claim 8, characterized in that: The robustness criteria for step S44 include the following conditions: Capacity standard: The absolute value of the deviation between actual capacity and planned capacity ≤ the preset allowable deviation; Cost standard: The absolute value of the deviation between the actual unit cost and the planned cost is less than or equal to the preset allowable deviation; Quality standard: Actual quality compliance rate ≥ Planned quality compliance rate.

Images