A real-time production collaborative processing method for supply chain platform

Abstract

The present invention discloses a real-time production collaborative processing method for a supply chain platform, which relates to the technical field of enterprise production and business management, and includes the following steps: S1, data collection and integration; S2, real-time demand forecasting; S3, dynamic configuration of production resources; S4, intelligent production scheduling; S5, abnormality detection and early warning; S6, collaborative production adjustment; S7, real-time feedback of production status; S8, adaptive production plan update; S9, real-time inventory management. The real-time production collaborative processing method for a supply chain platform realizes efficient dynamic adjustment in demand forecasting, resource allocation, and production task scheduling, can learn and optimize production plans in real time, dynamically adjust resource allocation, and automatically trigger production adjustment mechanisms through collaborative platforms when abnormalities or demand changes are detected, so as to ensure the efficiency and stability of the production process, significantly improve the flexibility and response speed of production scheduling, and realize the optimal configuration of production resources.

Description

Technical Field

[0001] The present invention relates to the technical field of enterprise production and business management, and in particular to a real-time production collaborative processing method for a supply chain platform. Background Art

[0002] In modern supply chain management, efficient coordination and dynamic adjustment of production are the key to ensuring that enterprises maintain their advantages in market competition. However, existing supply chain platforms usually rely on static production scheduling and resource allocation schemes, which makes it difficult for them to respond quickly when faced with rapidly changing market demands, sudden production anomalies and supply chain disruptions. This lagging response mechanism leads to low production efficiency, serious waste of resources, and may even lead to order delays or cancellations. In addition, traditional production scheduling systems are usually based on fixed rules or simple optimization algorithms, lack flexibility and adaptability, and cannot achieve real-time dynamic adjustments in complex and changing production environments.

[0003] In order to solve the above problems, we proposed a real-time production collaborative processing method for supply chain platform, which can achieve efficient dynamic adjustment in demand forecasting, resource allocation and production task scheduling. Summary of the invention

[0004] In view of the deficiencies in the prior art, the present invention provides a real-time production collaborative processing method for a supply chain platform to solve the problems raised in the above background technology.

[0005] To achieve the above object, the present invention provides the following technical solution: a real-time production collaborative processing method for a supply chain platform, comprising the following steps:

[0006] S1, data collection and integration;

[0007] Collect real-time data from all links of the supply chain and integrate the data into the supply chain platform;

[0008] The data includes the working status of the production line, the inventory of raw materials, order requirements, and order logistics information;

[0009] S2, real-time demand forecast;

[0010] Using historical data and current order demand, deep learning models are applied to predict future production demand;

[0011] S3, dynamic configuration of production resources;

[0012] Based on the results of demand forecasting, a real-time resource optimization model is used to dynamically configure production resources;

[0013] The real-time resource optimization model optimizes the configuration of production resources based on multi-dimensional factors such as production bottlenecks, equipment status, and human resources;

[0014] S4, intelligent production scheduling;

[0015] Use reinforcement learning algorithms to dynamically adjust the work order and task allocation of the production line;

[0016] S5, anomaly detection and early warning;

[0017] Monitor the status of each production link in real time, detect equipment failure, abnormal production or quality problems by monitoring sensor data and production logs, and give early warnings before problems occur;

[0018] S6, collaborative production adjustment;

[0019] When anomalies or changes in demand are detected, the production adjustment mechanism is automatically triggered through the collaborative platform;

[0020] Adjustment mechanisms include reconfiguration of production lines, adjustment of raw material supplies, and dynamic updating of order priorities;

[0021] S7, real-time feedback of production status;

[0022] During the production execution process, the production status is fed back to each link of the supply chain in real time;

[0023] S8, adaptive production plan update;

[0024] According to the real-time data feedback in the production process, the current production plan is dynamically adjusted based on the adaptive plan update algorithm;

[0025] S9. Real-time inventory management;

[0026] Real-time management of raw materials and finished product inventory is required during the production process.

[0027] To further optimize the technical solution, in step S2, the deep learning model is combined with time series analysis and dynamic regression to generate accurate demand forecasting results, the production plan is planned in advance based on the forecasting results of the deep learning model, and the production rhythm is adjusted in time when demand fluctuates.

[0028] To further optimize the technical solution, the deep learning model includes:

[0029] Input layer: Input data includes time series data and dynamic data of upstream and downstream supply chains;

[0030] Time series encoding layer: Use the long short-term memory network (LSTM) to capture the temporal dependencies in historical demand data. The formula is as follows:

[0031]

[0032] in, It's time The input data, is the current hidden state, is the hidden state at the previous moment, is the parameter set of the LSTM network;

[0033] Dynamic regression layer: The dynamic data of the supply chain is regressed through a generalized linear model (GLM) to predict short-term demand changes. The formula is as follows:

[0034]

[0035] in, It's in time The demand forecast value, It's the past Dynamic data at a time point, and is the regression coefficient, is the noise term;

[0036] Fusion layer: The outputs of the time series encoding layer and the dynamic regression layer are fused to generate the final prediction value. The fusion is completed using a fully connected network (FCN). The fusion formula is as follows:

[0037]

[0038] in, is the final prediction value after fusion, is the parameter set of FCN;

[0039] Output layer: Through a linear transformation, the fused results are converted into specific order demand prediction values.

[0040] To further optimize the technical solution, the deep learning model includes the following process when making predictions:

[0041] Data preprocessing: preprocess the input data, including standardization of time series data and feature extraction of dynamic data;

[0042] Model training: Use historical data to train deep learning models;

[0043] Real-time prediction: In actual applications, the deep learning model receives real-time updated data and uses trained parameters to make demand predictions;

[0044] Model update: Deep learning models have adaptive learning capabilities and can be updated online based on the latest data.

[0045] To further optimize the technical solution, in step S3, the real-time resource optimization model satisfies the following constraints during resource allocation:

[0046] Human Constraints:

[0047] in, For time The amount of manpower deployed at the time, For minimum manpower requirements;

[0048] Device constraints:

[0049] in, For time The amount of equipment used at For the minimum equipment requirements;

[0050] Material constraints:

[0051] in, For time The material configuration amount, Minimum material requirements;

[0052] To achieve optimal resource allocation, the real-time resource optimization model uses a dynamic scheduling algorithm that dynamically adjusts resource allocation based on demand forecast results and real-time data using the following formula:

[0053]

[0054] in, It's time The optimal resource allocation when , , is the weight coefficient of manpower, equipment and materials, is the bottleneck adjustment factor, is the severity of the production bottleneck;

[0055] Through the feedback control mechanism, the effect of resource allocation is monitored in real time and adjusted. The feedback formula is as follows:

[0056]

[0057] in, is the resource adjustment amount, is the target production efficiency, is the current production efficiency, is the feedback coefficient.

[0058] To further optimize the technical solution, the real-time resource optimization model includes the following processes during resource allocation:

[0059] Data input: The model first receives the prediction results from the deep learning model in step S2 and the real-time data of the current production line;

[0060] Resource allocation calculation: The model uses constraints to calculate the current time point The resource allocation amount is used to generate the optimal resource allocation plan ;

[0061] Real-time feedback adjustment: Through the feedback control mechanism, the model monitors production efficiency in real time and target efficiency the gap between;

[0062] Continuous optimization: Continuously optimize weight coefficients based on historical data and real-time feedback and feedback coefficient .

[0063] To further optimize the technical solution, in step S4, the process of the reinforcement learning algorithm is as follows:

[0064] Environment initialization: define the current state of the production line , including information on resource allocation, manpower, equipment status, material inventory, production task queue, etc.;

[0065] Action space definition: define scheduling actions , including adjusting the sequence of production tasks, changing resource allocation, and reallocating manpower or equipment;

[0066] Reward function definition: Setting the reward function , used to evaluate the effect of each action;

[0067] Strategy selection: Select the strategy of reinforcement learning algorithm, including Q learning and deep Q network (DQN);

[0068] State transition: Execute the selected action , the production system changes from the current state Transition to a new state , and receive corresponding rewards ;

[0069] Update strategy: Update the strategy based on the result of the current action;

[0070] Explore and exploit: During the learning process, try new scheduling strategies and apply existing optimal strategies;

[0071] Termination conditions: Set the termination conditions for scheduling optimization, such as when the predetermined production efficiency is reached, a fixed number of time steps have passed, or all tasks are completed, the algorithm stops running;

[0072] Online learning and real-time optimization: continuously acquire new data from the production process and continuously optimize strategies;

[0073] Apply scheduling results: When the strategy optimization is completed, the final scheduling results are applied to the production line to adjust the production task sequence and resource allocation.

[0074] To further optimize the technical solution, in step S6, the adjustment mechanism includes:

[0075] The collaborative platform is integrated with the supply chain platform. When changes in demand are detected, the collaborative platform automatically adjusts the supply plan of raw materials. If some materials are in short supply, the collaborative platform will dynamically find alternative materials or adjust the production plan.

[0076] When demand changes or abnormal situations occur, the collaborative platform automatically updates the order priority according to the new conditions. After the priority is updated, production tasks are reallocated to ensure that high-priority orders can be completed on time;

[0077] The collaborative platform uses distributed computing capabilities to process large amounts of data and complex task rescheduling calculations, distributes tasks among multiple production departments, and ensures optimal utilization of resources and continuity of production.

[0078] To further optimize the technical solution, in step S8, the adaptive plan update algorithm automatically updates the production sequence, adjusts resource allocation, and recalculates the production schedule based on real-time data input.

[0079] To further optimize this technical solution, the method also includes performance analysis of the collaborative processing of production tasks, using big data analysis technology to identify bottlenecks and inefficient links in production and make optimization suggestions. The analysis results are not only used to optimize the current production process, but also input into various models to improve future production scheduling and resource allocation strategies.

[0080] Compared with the prior art, the present invention provides a real-time production collaborative processing method for a supply chain platform, which has the following beneficial effects:

[0081] This real-time production collaborative processing method for the supply chain platform achieves efficient dynamic adjustment in demand forecasting, resource allocation, and production task scheduling. It can learn and optimize production plans in real time, dynamically adjust resource allocation, and automatically trigger the production adjustment mechanism through the collaborative platform when anomalies or demand changes are detected, thereby ensuring the efficiency and stability of the production process, significantly improving the flexibility and response speed of production scheduling, effectively responding to rapidly changing market demands and sudden production anomalies, and achieving the optimal allocation of production resources, thereby improving overall production efficiency, reducing resource waste, and ensuring on-time delivery of orders. BRIEF DESCRIPTION OF THE DRAWINGS

[0082] Figure 1 A schematic diagram of a flow chart of a real-time production collaborative processing method for a supply chain platform proposed by the present invention;

[0083] Figure 2 A schematic diagram of a deep learning model in a real-time production collaborative processing method for a supply chain platform proposed by the present invention;

[0084] Figure 3 A flow chart of a real-time resource optimization model in a real-time production collaborative processing method for a supply chain platform proposed by the present invention;

[0085] Figure 4 This is a flow chart of a reinforcement learning algorithm in a real-time production collaborative processing method for a supply chain platform proposed by the present invention. DETAILED DESCRIPTION

[0086] The following will be combined with the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention. Example

[0087] See also Figure 1 , a real-time production collaborative processing method for a supply chain platform, comprising the following steps:

[0088] S1, data collection and integration;

[0089] Collect real-time data from all links of the supply chain and integrate the data into the supply chain platform.

[0090] The data includes the working status of the production line, inventory status of raw materials, order requirements, and order logistics information.

[0091] In this embodiment, by integrating these data into the platform, information synchronization between modules can be ensured and collaboration problems caused by information islands can be reduced.

[0092] S2, real-time demand forecast;

[0093] Using historical data and current order demand, deep learning models are applied to predict future production demand.

[0094] In this embodiment, the deep learning model combines time series analysis and dynamic regression to generate accurate demand forecast results. The production plan is planned in advance based on the prediction results of the deep learning model, and the production rhythm is adjusted in time when demand fluctuates.

[0095] like Figure 2 As shown, the deep learning model includes:

[0096] Input layer: The input data includes time series data (historical demand, seasonal factors, etc.) and dynamic data of the upstream and downstream supply chains (such as inventory levels, transportation time, market trends, etc.).

[0097] Time series encoding layer: This layer uses a long short-term memory network (LSTM) to capture the temporal dependencies in historical demand data. The formula is as follows:

[0098]

[0099] in, It's time The input data, is the current hidden state, is the hidden state at the previous moment, is the parameter set of the LSTM network;

[0100] Dynamic regression layer: This layer uses a generalized linear model (GLM) to perform regression analysis on the dynamic data of the supply chain to predict short-term demand changes. The formula is as follows:

[0101]

[0102] in, It's in time The demand forecast value, It's the past Dynamic data at a time point, and is the regression coefficient, is the noise term.

[0103] Fusion layer: The output of the time series encoding layer and the dynamic regression layer are fused to generate the final prediction value. Here, a fully connected network (FCN) is used to complete this fusion:

[0104]

[0105] in, is the final prediction value after fusion, is the parameter set of FCN;

[0106] Output layer: The output layer converts the fused results into specific demand forecast values ​​through a linear transformation.

[0107] The deep learning model includes the following processes when making predictions:

[0108] Data preprocessing: preprocess the input data, including standardization of time series data and feature extraction of dynamic data. The preprocessed data is used as the input of the model to improve the accuracy of the prediction.

[0109] Model training: Use historical data to train the deep learning model. The LSTM of the time series encoding layer captures the long-term dependencies of the data, and the GLM of the dynamic regression layer analyzes the short-term fluctuations of dynamic data. During the training process, the model automatically adjusts the parameters. , and the regression coefficient , in order to minimize the prediction error.

[0110] Real-time prediction: In actual applications, the deep learning model receives real-time updated data and uses the trained parameters to make demand predictions. The output of the model For the current time point The forecast demand value can be used to guide the adjustment of production plans and resource allocation.

[0111] Model update: The DDPN model also has adaptive learning capabilities and can be updated online based on the latest data. In this way, the model can continuously optimize itself to cope with changes in the supply chain environment and demand fluctuations.

[0112] The combined use of LSTM and GLM can capture both long-term and short-term demand changes. This multi-level forecasting approach enables the model to provide more accurate and reliable forecast results when facing a complex and dynamic supply chain environment.

[0113] S3, dynamic configuration of production resources;

[0114] Based on the results of demand forecasting, a real-time resource optimization model is used to dynamically configure production resources.

[0115] The real-time resource optimization model optimizes the configuration of production resources based on multi-dimensional factors such as production bottlenecks, equipment status, and human resources.

[0116] In this embodiment, the real-time resource optimization model satisfies the following constraints during resource allocation:

[0117] Human Constraints:

[0118] in, For time The amount of manpower deployed at the time, For minimum manpower requirements;

[0119] Device constraints:

[0120] in, For time The amount of equipment used at For the minimum equipment requirements;

[0121] Material constraints:

[0122] in, For time The material configuration amount, Minimum material requirements;

[0123] To achieve optimal resource allocation, the real-time resource optimization model uses a dynamic scheduling algorithm that dynamically adjusts resource allocation based on demand forecast results and real-time data using the following formula:

[0124]

[0125] in, It's time The optimal resource allocation when , , is the weight coefficient of manpower, equipment and materials, is the bottleneck adjustment factor, is the severity of the production bottleneck;

[0126] Through the feedback control mechanism, the effect of resource allocation is monitored in real time and adjusted. The feedback formula is as follows:

[0127]

[0128] in, is the resource adjustment amount, is the target production efficiency, is the current production efficiency, is the feedback coefficient.

[0129] like Figure 3 As shown, the real-time resource optimization model includes the following processes when allocating resources:

[0130] Data input:

[0131] The model first receives the prediction results from the deep learning model in step S2, as well as the real-time data of the current production line, including information such as manpower allocation, equipment status, material inventory, and production bottlenecks. All input data will be standardized and input into the model.

[0132] Resource allocation calculation:

[0133] The model uses constraints to calculate the current time point The dynamic scheduling algorithm dynamically adjusts the resource allocation weight according to demand fluctuations and real-time production conditions to generate the optimal resource allocation plan. .

[0134] Real-time feedback adjustment:

[0135] Through the feedback control mechanism, the model monitors production efficiency in real time and target efficiency If the current production efficiency does not meet expectations, the model will automatically adjust resource allocation and calculate the new resource adjustment amount. , and applied immediately in the production process.

[0136] Continuous Optimization:

[0137] The model has self-learning capabilities and can continuously optimize weight coefficients based on historical data and real-time feedback. and feedback coefficient , in order to improve the accuracy and response speed of future resource allocation.

[0138] By introducing the severity of production bottlenecks into resource allocation decisions and using feedback control mechanisms to adjust resource allocation in real time, we can flexibly respond to various resource allocation needs in complex and changing production environments and ensure the efficiency and stability of the production process.

[0139] S4, intelligent production scheduling;

[0140] Use reinforcement learning algorithms to dynamically adjust the work order and task allocation of the production line.

[0141] like Figure 4 As shown, in this embodiment, in step S4, the process of the reinforcement learning algorithm is as follows:

[0142] Environment initialization:

[0143] Define the current status of the production line , including information such as resource allocation, manpower, equipment status, material inventory, production task queue, etc. After the environment is initialized, the production scheduling system will use this as the initial state for optimization.

[0144] Action space definition:

[0145] Define possible dispatch actions , including adjusting the order of production tasks, changing resource allocation, reallocating manpower or equipment, etc. These actions are decisions that the system can choose to execute.

[0146] Reward function definition:

[0147] Setting the reward function , which is used to evaluate the effect of each action. The reward function is calculated based on factors such as production efficiency, resource utilization, and production line stability.

[0148] Strategy selection:

[0149] Select the strategy of reinforcement learning algorithm, including Q learning and deep Q network (DQN). The purpose of the strategy is to choose the best action in a given state to maximize the long-term cumulative reward.

[0150] State transfer:

[0151] Perform the selected action , the production system changes from the current state Transition to a new state , and receive corresponding rewards .

[0152] Update strategy:

[0153] Update the strategy based on the result of the current action. A common approach is to adjust the parameters of the strategy by maximizing the cumulative reward so that the system can choose better actions in future decisions.

[0154] Explore and Exploit:

[0155] During the learning process, new scheduling strategies are tried and the existing optimal strategy is applied. This is usually done by - Greedy strategy to achieve: with probability Choose random actions (exploration) with probability Select the current best action (utilization).

[0156] Termination conditions:

[0157] The termination conditions of scheduling optimization are set, such as when the predetermined production efficiency is reached, a fixed number of time steps have passed, or all tasks are completed, the algorithm stops running.

[0158] Online learning and real-time optimization:

[0159] The reinforcement learning algorithm has the ability to learn online, and can continuously obtain new data from the production process and continuously optimize the strategy. Whenever there is a new emergency or production interruption, the system can dynamically adjust the scheduling strategy and respond quickly.

[0160] Application scheduling results:

[0161] When the strategy optimization is completed, the final scheduling results are applied to the production line to adjust the production task sequence and resource allocation to achieve the optimal production scheduling.

[0162] S5, anomaly detection and early warning;

[0163] Monitor the status of each production link in real time, detect equipment failure, abnormal production or quality problems by monitoring sensor data and production logs, and give early warnings before problems occur.

[0164] In this embodiment, when the status of each production link is monitored in real time, the monitoring threshold and strategy are automatically adjusted as the production environment changes.

[0165] S6, collaborative production adjustment;

[0166] When anomalies or changes in demand are detected, the production adjustment mechanism is automatically triggered through the collaborative platform.

[0167] Adjustment mechanisms include reconfiguration of production lines, adjustment of raw material supply, and dynamic updating of order priorities.

[0168] In this embodiment, the adjustment mechanism includes:

[0169] The collaborative platform is integrated with the supply chain platform. When changes in demand are detected, the collaborative platform automatically adjusts the supply plan of raw materials. If some materials are in short supply, the collaborative platform will dynamically find alternative materials or adjust the production plan.

[0170] When demand changes or abnormal situations occur, the collaborative platform automatically updates the order priority according to the new conditions. After the priority is updated, production tasks are reallocated to ensure that high-priority orders can be completed on time;

[0171] The collaborative platform uses distributed computing capabilities to process large amounts of data and complex task rescheduling calculations, distributes tasks among multiple production departments, and ensures optimal utilization of resources and continuity of production.

[0172] S7, real-time feedback of production status;

[0173] During the production execution process, the production status is fed back to each link of the supply chain in real time.

[0174] In this embodiment, by using a high-frequency and low-latency communication protocol (such as MQTT), it is ensured that production data can be transmitted to the decision-making layer and related modules with minimal delay, and adjustments and optimizations are made based on real-time information.

[0175] S8, adaptive production plan update;

[0176] According to the real-time data feedback in the production process, the current production plan is dynamically adjusted based on the adaptive plan update algorithm.

[0177] In this embodiment, the adaptive plan update algorithm automatically updates the production sequence, adjusts resource allocation, and recalculates the production schedule based on real-time data input.

[0178] S9. Real-time inventory management;

[0179] Real-time management of raw materials and finished product inventory is required during the production process.

[0180] In this embodiment, through intelligent inventory management, inventory levels are monitored in real time, and replenishment or delivery instructions are automatically triggered when necessary.

[0181] Embodiment 2:

[0182] Based on the real-time production collaborative processing method for a supply chain platform described in Example 1, the method also includes a performance analysis of the collaborative processing process of production tasks, using big data analysis technology to identify bottlenecks and inefficient links in production, and put forward optimization suggestions. The analysis results are not only used to optimize the current production process, but also input into various models to improve future production scheduling and resource allocation strategies.

[0183] The beneficial effects of the present invention are:

[0184] This real-time production collaborative processing method for the supply chain platform achieves efficient dynamic adjustment in demand forecasting, resource allocation, and production task scheduling. It can learn and optimize production plans in real time, dynamically adjust resource allocation, and automatically trigger the production adjustment mechanism through the collaborative platform when anomalies or demand changes are detected, thereby ensuring the efficiency and stability of the production process, significantly improving the flexibility and response speed of production scheduling, effectively responding to rapidly changing market demands and sudden production anomalies, and achieving the optimal allocation of production resources, thereby improving overall production efficiency, reducing resource waste, and ensuring on-time delivery of orders.

[0185] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, without contradiction.

[0186] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A real-time production collaborative processing method for a supply chain platform, characterized in that: The following steps are involved: S1, data collection and integration; Collect real-time data from all links of the supply chain and integrate the data into the supply chain platform; The data includes the working status of the production line, the inventory of raw materials, order requirements, and order logistics information; S2, real-time demand forecast; Using historical data and current order demand, deep learning models are applied to predict future production demand; S3, dynamic configuration of production resources; Based on the results of demand forecasting, a real-time resource optimization model is used to dynamically configure production resources; The real-time resource optimization model optimizes the configuration of production resources based on multi-dimensional factors such as production bottlenecks, equipment status, and human resources; The real-time resource optimization model meets the following constraints during resource allocation: Human Constraints: ; in, For time The amount of manpower deployed at the time, For minimum manpower requirements; Device constraints: ; in, For time The amount of equipment used at For the minimum equipment requirements; Material constraints: ; in, For time The material configuration amount, Minimum material requirements; To achieve optimal resource allocation, the real-time resource optimization model uses a dynamic scheduling algorithm that dynamically adjusts resource allocation based on demand forecast results and real-time data using the following formula: ;in, It's time The optimal resource allocation when , , is the weight coefficient of manpower, equipment and materials, is the bottleneck adjustment factor, is the severity of the production bottleneck; Through the feedback control mechanism, the effect of resource allocation is monitored in real time and adjusted. The feedback formula is as follows: ;in, is the resource adjustment amount, is the target production efficiency, is the current production efficiency, is the feedback coefficient; The real-time resource optimization model includes the following processes when allocating resources: Data input: The model first receives the prediction results from the deep learning model in step S2 and the real-time data of the current production line; Resource allocation calculation: The model uses constraints to calculate the current time point The resource allocation amount is used to generate the optimal resource allocation plan ; Real-time feedback adjustment: Through the feedback control mechanism, the model monitors production efficiency in real time and target efficiency the gap between; Continuous optimization: Continuously optimize weight coefficients based on historical data and real-time feedback , , and feedback coefficient ; S4, intelligent production scheduling; Use reinforcement learning algorithms to dynamically adjust the work order and task allocation of the production line; S5, anomaly detection and early warning; Monitor the status of each production link in real time, detect equipment failure, abnormal production or quality problems by monitoring sensor data and production logs, and give early warnings before problems occur; S6, collaborative production adjustment; When anomalies or changes in demand are detected, the production adjustment mechanism is automatically triggered through the collaborative platform; Adjustment mechanisms include reconfiguration of production lines, adjustment of raw material supplies, and dynamic updating of order priorities; S7, real-time feedback of production status; During the production execution process, the production status is fed back to each link of the supply chain in real time; S8, adaptive production plan update; According to the real-time data feedback in the production process, the current production plan is dynamically adjusted based on the adaptive plan update algorithm; S9. Real-time inventory management; Real-time management of raw materials and finished product inventory is required during the production process.

2. A real-time production collaborative processing method for a supply chain platform according to claim 1, characterized in that: In step S2, the deep learning model combines time series analysis and dynamic regression to generate accurate demand forecast results, plans production plans in advance based on the forecast results of the deep learning model, and adjusts the production rhythm in time when demand fluctuates.

3. A real-time production collaborative processing method for a supply chain platform according to claim 2, characterized in that: The deep learning model includes: Input layer: Input data includes time series data and dynamic data of upstream and downstream supply chains; Time series encoding layer: Use the long short-term memory network LSTM to capture the time dependency in historical demand data. The formula is as follows: ； in, It's time The input data, is the current hidden state, is the hidden state at the previous moment, is the parameter set of the LSTM network; Dynamic regression layer: The dynamic data of the supply chain is regressed through the generalized linear model GLM to predict short-term demand changes. The formula is as follows: ； in, It's in time The demand forecast value, It's the past Dynamic data at a time point, and is the regression coefficient, is the noise term; Fusion layer: The outputs of the time series encoding layer and the dynamic regression layer are fused to generate the final prediction value. The fusion is completed using the fully connected network FCN. The fusion formula is as follows: ； in, is the final prediction value after fusion, is the parameter set of FCN; Output layer: Through a linear transformation, the fused results are converted into specific order demand prediction values.

4. A real-time production collaborative processing method for a supply chain platform according to claim 3, characterized in that: The deep learning model includes the following processes when making predictions: Data preprocessing: preprocess the input data, including standardization of time series data and feature extraction of dynamic data; Model training: Use historical data to train deep learning models; Real-time prediction: In actual applications, the deep learning model receives real-time updated data and uses trained parameters to make demand predictions; Model update: Deep learning models have adaptive learning capabilities and can be updated online based on the latest data.

5. The real-time production collaborative processing method for a supply chain platform according to claim 1, characterized in that: In step S4, the process of the reinforcement learning algorithm is as follows: Environment initialization: define the current state of the production line , including information on resource allocation, manpower, equipment status, material inventory, and production task queues; Action space definition: define scheduling actions , including adjusting the sequence of production tasks, changing resource allocation, and reallocating manpower or equipment; Reward function definition: Setting the reward function , used to evaluate the effect of each action; Strategy selection: Select the strategy of reinforcement learning algorithm, including Q learning and deep Q network DQN; State transition: Execute the selected action , the production system changes from the current state Transition to a new state , and receive corresponding rewards ; Update strategy: Update the strategy based on the result of the current action; Explore and exploit: During the learning process, try new scheduling strategies and apply existing optimal strategies; Termination conditions: Set the termination conditions for scheduling optimization, such as when the predetermined production efficiency is reached, a fixed number of time steps have passed, or all tasks are completed, the algorithm stops running; Online learning and real-time optimization: continuously acquire new data from the production process and continuously optimize strategies; Apply scheduling results: When the strategy optimization is completed, the final scheduling results are applied to the production line to adjust the production task sequence and resource allocation.

6. A real-time production collaborative processing method for a supply chain platform according to claim 1, characterized in that: In step S6, the adjustment mechanism includes: The collaborative platform is integrated with the supply chain platform. When changes in demand are detected, the collaborative platform automatically adjusts the supply plan of raw materials. If some materials are in short supply, the collaborative platform will dynamically find alternative materials or adjust the production plan. When demand changes or abnormal situations occur, the collaborative platform automatically updates the order priority according to the new conditions. After the priority is updated, production tasks are reallocated to ensure that high-priority orders can be completed on time; The collaborative platform uses distributed computing capabilities to process large amounts of data and complex task rescheduling calculations, distributes tasks among multiple production departments, and ensures optimal utilization of resources and continuity of production.

7. A real-time production collaborative processing method for a supply chain platform according to claim 1, characterized in that: In step S8, the adaptive plan update algorithm automatically updates the production sequence, adjusts resource allocation, and recalculates the production schedule based on real-time data input.

8. The real-time production collaborative processing method for a supply chain platform according to claim 1, characterized in that: The method also includes performance analysis of the collaborative processing of production tasks. It uses big data analysis technology to identify bottlenecks and inefficient links in production and make optimization suggestions. The analysis results are not only used to optimize the current production process, but also input into various models to improve future production scheduling and resource allocation strategies.

Images