MES data driving-based shoe production takt intelligent scheduling and control method

Abstract

The invention discloses a shoe production takt intelligent scheduling and control method based on MES data driving. The core of the method is that the queue length, waiting time and workpiece color matching information of each process node are collected in real time; dynamically calculating a process node comprehensive pressure value fused with the color mismatch degree; identifying the maximum pressure node of the whole line; a scheduling instruction is generated according to the color mismatch degree and comprises the steps of commanding upstream throttling and downstream dredging and triggering a priority pairing instruction when the color mismatch degree exceeds the standard; visual guide execution is carried out through a station self-adaptive metronome; and meanwhile, the feeding rhythm is globally regulated and controlled based on tail end pressure The production line is regarded as a dynamic pressure system, conversion from static planning to dynamic data driving is achieved, dynamic bottlenecks are effectively eliminated, inventory of products in process and'isolated shoes' is greatly reduced, and production efficiency and order delivery punctuality rate are remarkably improved.

Description

Technical Field

[0001] This invention belongs to the field of industrial automation and intelligent manufacturing technology, specifically a method for intelligent scheduling and control of footwear production cycle based on MES data. Background Technology

[0002] Footwear production lines typically consist of multiple processes linked together, including cutting, sewing, molding, and quality inspection. Currently, the mainstream production cycle scheduling and control methods in the industry mainly rely on the following approaches:

[0003] 1. Push-based production scheduling based on fixed plans:

[0004] This is the most traditional and widely used method. Based on monthly or weekly production plans and standard working hours, the production management department creates detailed daily production schedules for each production line, specifying output targets for each process within a specific time period. During production, materials and semi-finished products are "pushed" to the next process according to the predetermined plan. This method is rigid and cannot cope with dynamic changes on the production floor. When a process experiences a slight delay due to equipment malfunctions, material problems, or fluctuations in personnel efficiency, this delay accumulates and amplifies along the production line, causing subsequent processes to idle or upstream processes to accumulate work-in-process inventory, forming a hidden bottleneck. Team leaders often only intervene when the problem has already manifested (such as severe backlog), resulting in a delayed response and overall low production efficiency and high work-in-process inventory.

[0005] 2. Static scheduling based on simple MES data monitoring

[0006] With the widespread adoption of Manufacturing Execution Systems (MES), some solutions attempt to utilize data such as output and working hours collected by MES for post-production statistical analysis and adjust future production plans accordingly. Others involve setting simple threshold alarms in the system to alert users when output falls below a set value. However, this approach is inherently static and reactive. It relies on historical data analysis and cannot provide real-time, forward-looking dynamic control of the production line. Its alarm mechanisms typically only notify users after a problem occurs, failing to automatically implement corrective measures in the early stages of a problem, thus lacking intelligence.

[0007] Especially in footwear production, there is a unique technical challenge—the "matching" problem. Footwear needs to be produced and packaged in pairs, left and right. In all the existing scheduling methods mentioned above, the system treats each workpiece as an independent entity, focusing only on its quantity and location, without sensing its "pairing" status. This means that even slight asynchrony in upstream processes can result in a large number of "lone shoes" (unmatched single shoes) downstream. These "lone shoes" accumulate at the end of the production line, not only tying up significant inventory and severely impacting order delivery cycles, but also causing chaos in production site management. Existing technology is largely ineffective in addressing this problem, relying solely on manual sorting and scheduling at the end, which is extremely inefficient.

[0008] In summary, existing footwear production scheduling methods are either rigid and lagging or slow to respond, generally lacking the ability to perceive and intelligently respond to the dynamic flow state within the production line, and failing to address the unique "color matching" challenges of the footwear industry. Therefore, there is an urgent need in this field for an innovative scheduling and control method capable of real-time perception of production flow status, intelligent dynamic adjustment of the production rhythm, and precise resolution of the unique problems in footwear production. Summary of the Invention

[0009] The purpose of this invention is to provide a method for intelligent scheduling and control of footwear production cycle based on MES data. This invention treats the production line as a dynamic pressure system, realizing the transformation from static planning to dynamic data-driven, effectively eliminating dynamic bottlenecks, significantly reducing work-in-process and "lone shoes" inventory, and significantly improving production efficiency and order delivery on-time rate.

[0010] The technical solution adopted in this invention is as follows:

[0011] A method for intelligent scheduling and control of footwear production cycle based on MES data-driven approach, characterized by the following steps:

[0012] Step S1: Collect production status data of each process node on the production line in real time. The production status data includes at least the length of the work-in-process queue in front of each process node and the waiting time of each workpiece in the queue.

[0013] Step S2: Based on the production status data, dynamically calculate the real-time pressure value of each process node;

[0014] Step S3: Identify the process node with the highest real-time pressure value across the entire line and record it as the maximum pressure node;

[0015] Step S4: Based on the location and pressure value of the maximum pressure node, generate a scheduling control instruction, wherein the scheduling control instruction includes at least: issuing a first instruction to the direct upstream process node of the maximum pressure node, the first instruction being used to reduce the production cycle time of the upstream process node; and issuing a second instruction to the downstream process node of the maximum pressure node, the second instruction being used to accelerate the production cycle time of the downstream process node.

[0016] Step S5: Send the scheduling control command to the terminal display device of the corresponding process node for visual prompts to guide the operator to execute it.

[0017] Preferably, in step S2, the real-time pressure value Pi(t) is calculated using the following formula: Where Pi(t) is the pressure value of process node i at time t, Li(t) is the queue length of process node i at time t, Ti(t) is the average waiting time of the workpiece in the queue of process node i at time t, and α and β are preset weighting coefficients.

[0018] Preferably, in step S4, the rate of decrease in the cycle time of the first instruction is positively correlated with the rate by which the pressure value of the maximum pressure node exceeds the average pressure value of the entire line; and / or, the rate of increase in the cycle time of the second instruction is positively correlated with the rate by which the pressure value of the downstream process node of the maximum pressure node is lower than the average pressure value of the entire line.

[0019] Preferably, the method further includes a step of controlling the feeding cycle of the production line:

[0020] Step S6: Monitor the pressure value of the final process node at the end of the production line in real time;

[0021] Step S7: Dynamically adjust the feeding cycle at the beginning of the production line according to the pressure value of the final process node; when the pressure value of the final process node is lower than the first threshold, increase the feeding cycle; when the pressure value of the final process node is higher than the second threshold, decrease or suspend the feeding cycle.

[0022] Preferably, different instruction states are distinguished by changing the background color of the terminal display device, wherein: green background represents normal rhythm; yellow background represents acceleration instruction; and red background represents deceleration or pause instruction.

[0023] Preferably, the visual prompts also include dynamically displaying the theoretical operating rhythm of the current process on the terminal display device in the form of a metronome or countdown timer, so as to guide the operator's operating rhythm.

[0024] Preferably, in step S1, the production status data further includes color code and size information determined based on the unique identification information of each workpiece; in step S2, the formula for calculating the real-time pressure value Pi(t) is: Where Ci(t) is the color mismatch degree of process node i at time t, and its value is the ratio of the number of workpieces in the queue that cannot be matched based on the color code and size information to the total length of the queue Li(t); γ is a preset weight coefficient.

[0025] Preferably, when the system identifies a certain process node as the node with the greatest pressure and its color mismatch Ci(t) exceeds the third threshold, the scheduling control instruction generated in step S4 further includes: issuing a third instruction to the upstream process node of the node, wherein the third instruction is used to optimize the feeding or processing sequence and prioritize the processing of workpieces that can be paired with workpieces that cannot be paired in the current queue.

[0026] A footwear production cycle intelligent scheduling and control system for implementing the method includes: a sensing module configured to collect production status data of each process node in real time; a decision module configured to calculate the real-time pressure value of each node, identify the node with the maximum pressure, and generate scheduling control instructions; and an execution module including terminal display devices set at each process node, configured to receive and visualize the scheduling control instructions.

[0027] A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method.

[0028] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are:

[0029] Traditional scheduling is like driving in fog, relying on experience to guess road conditions. This invention, however, installs a high-precision "pressure sensor network" and "autonomous driving system" on the production line. By calculating the comprehensive pressure value of each process node in real time, the system can detect the slightest blockage (bottleneck) within seconds, much like a human nerve reflex, and react immediately: instructing upstream to "throttle" and urging downstream to "dredge." This results in a significant improvement in production efficiency and a substantial reduction in production cycle time. More importantly, the unique "color mismatch" index allows the system to accurately identify "lone shoe" problems midway through the production line and proactively allocate resources for matching, fundamentally solving the core pain point of "mismatched colors" in the footwear industry, and significantly reducing related inventory and delivery delays.

[0030] At a deeper level, this invention brings about an unexpected cognitive revolution: it transforms the traditional concept of a "bottleneck" that needs to be eliminated into a "compass" driving global optimization. The system no longer views bottlenecks as enemies, but instead utilizes the strongest signal—the "maximum pressure node"—as the core basis for coordinating the scheduling of the entire production line. This approach of "utilizing bottlenecks rather than fighting them" achieves a higher level of dynamic load balancing—it doesn't pursue rigid theoretical equal beat rates, but rather pursues flow matching based on real-time load, enabling the production line to maintain optimal output even during fluctuations.

[0031] This transformation also fostered unexpected synergies. The system handled all the complex, high-frequency data calculations and decisions, while delivering instructions to operators through an extremely intuitive visual interface (colors, metronome). This created an optimal human-machine division of labor: the machine handled its strengths in rapid calculations and global optimization, while humans leveraged their flexibility and on-site judgment. The result was not only increased efficiency but also a complete liberation of frontline managers from the burden of daily scheduling, allowing them to focus on higher-value tasks such as process improvement, quality control, and personnel training, leading to a qualitative leap in management effectiveness. Attached Figure Description

[0032] Figure 1 This is a schematic diagram of the hardware architecture of the present invention;

[0033] Figure 2 This is a schematic diagram of the layout of the present invention;

[0034] Figure 3 This is a flowchart of the control method of the present invention;

[0035] Figure 4 This is a hardware component framework diagram of the present invention. Detailed Implementation

[0036] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0037] See Figure 1 , 2 4. This invention relates to an intelligent scheduling and control method for footwear production rhythm based on MES data-driven methods. The hardware foundation of the control method system consists of a perception layer, a network layer, a decision layer, and an execution layer.

[0038] The sensing layer includes RFID tags and RFID readers. Each basket or workpiece box carrying footwear components is fitted with an UHF passive RFID tag. This tag has a unique TID number and the following information is written into the user's memory or associated with the MES database via its ID: batch number, style code, color code, size, and foot type (L / R). The tag must be wear-resistant, dustproof, and waterproof.

[0039] RFID readers are installed at the entrance and exit of each process node, with one fixed UHF RFID reader deployed at each. The readers are powered by PoE, and their antennas are precisely oriented to ensure stable reading of workpieces as they pass on the conveyor belt. For example, the entrance reader records that a workpiece has "entered the process queue," and the exit reader records that a workpiece has "completed the process."

[0040] The network layer is used for data transmission. Field readers connect to the factory's industrial Ethernet network via industrial switches. The readers send the collected data, including Tag_ID, reader's own serial number, timestamp, and RSSI signal strength, to the industrial gateway in the field. The gateway performs preliminary data filtering, such as noise reduction and deduplication, and adds a precise timestamp before publishing it to a specified topic in JSON format via the MQTT protocol.

[0041] The decision-making layer, the "brain" of the system, consists of one or more industrial servers. These servers are equipped with message queues to receive massive amounts of sensor data, time-series databases for efficient storage and retrieval of timestamped production events, and microservices for core scheduling algorithms.

[0042] The execution layer includes a display screen or device control unit. At each workstation, a low-power e-paper display screen is deployed as an "adaptive metronome." This display screen connects to the server via Wi-Fi to receive and visualize scheduling instructions.

[0043] See Figure 3 Specific control methods include:

[0044] Step S1: Real-time data acquisition and status calculation.

[0045] This step is the sensing link that initiates this method. Its core mission is to act as the "sensory nerves" of the entire intelligent scheduling system, continuously transforming physical events occurring on the production line into a set of structured state data that can be used for quantitative analysis. This step is not a one-time action, but a data pipeline that never stops during system operation.

[0046] The entire process begins with the most basic physical interaction. When a workpiece carrier with an RFID tag enters or leaves a specific area of ​​a process node via a conveyor belt, the RFID reader deployed in that area is triggered. This trigger generates a primitive "atomic event" that contains at least three key pieces of information: "who" (the tag's unique ID), "where" (the reader's physical identifier), and "when" (a timestamp accurate to milliseconds).

[0047] These raw, massive amounts of atomic events are immediately sent to a device called an "edge gateway" for initial processing. The gateway's primary task is to perform "debouncing" filtering, eliminating duplicate readings of the same tag due to signal reflections and other reasons, ensuring that each valid entry / exit action is recorded only once. Next, the system assigns crucial business semantics to these purified events. It translates the physical "reader number" into logical "process node" and "event type" by querying a pre-defined mapping table (e.g., reader "B1" corresponds to the business meaning of "workpiece entering the sewing machine process queue"). At this point, the chaotic physical signals are transformed into a "process event flow" with clear business meaning.

[0048] However, discrete events cannot be directly used to assess the health of the production line. Therefore, the system's core scheduler performs a "snapshot" analysis of all event streams at a fixed tick to calculate two core status metrics: real-time queue length (Li(t)) and average waiting time (Ti(t)).

[0049] The calculation logic for the queue length (Li(t)) is simple yet rigorous: For any monitored process node, the system performs a logical roll call. It identifies all workpieces that were recorded as "entering" the node before the current snapshot time t, but for which no record indicates they have "left." The total number of these workpieces is the queue length Li(t) of that node at time t. This precisely answers the question, "How many workpieces are currently queuing in front of this process?"

[0050] Based on this, the calculation of the average waiting time (Ti(t)) becomes straightforward. The system calculates the waiting time for each workpiece in the queue (current time t minus its "entry" timestamp), and then averages the waiting times of all workpieces. This value effectively reflects the flow rate of the queue; a continuously increasing Ti(t) is a strong signal of poor flow.

[0051] For the specific scenario of footwear production, the system performs a key enhancement at this snapshot moment: it scans the "identity information" (color code, size) of each item in the queue and attempts to pair left and right shoes within the queue. The ratio of the number of "lone shoes" that cannot be successfully paired to the total length of the queue is defined as the "color mismatch (Ci(t))". The introduction of this innovative indicator allows the system to not only perceive the accumulation of "quantity" but also to discern the imbalance of "quality" (whether they are complete sets), which is key to solving the bottleneck problems unique to footwear production.

[0052] Finally, in step S1, for each process node i at time t, a set of standardized state vectors is output: [Li(t), Ti(t), Ci(t)]. This set of vectors is no longer an isolated historical record, but rather a description of the real-time dynamic and continuous state variables of the production line. They are like high-precision instruments installed on each process, continuously transmitting raw data of "pressure" to the downstream decision engine (step S2), thereby driving the entire intelligent scheduling system to make precise responses.

[0053] Step S2: Dynamically calculate the real-time comprehensive pressure value Pi(t) of the process node.

[0054] This step takes the raw state data output from S1 and transforms it into a single, comparable metric that can uniformly measure the "congestion urgency" of each process node. The inevitable requirement of transforming multi-dimensional states into a single comparable metric: Step S1 provides each process node with three (or two) key state quantities at time t: queue length Li(t), average waiting time Ti(t), and color mismatch Ci(t). However, these three metrics, with different dimensions and numerical ranges, cannot be directly compared. Managers cannot answer whether a queue of length 5 or a queue with an average waiting time of 10 minutes is more urgent. Therefore, a mechanism is needed to integrate this information into a standardized, cross-process comparable "stress value," providing a unique and clear basis for global scheduling decisions. The stress value fusion calculation and quantification step S2 is automatically executed periodically by the scheduling algorithm, and its core is a weighted calculation formula: The formula works as follows:

[0055] The algorithm receives state vectors [Li(t), Ti(t), Ci(t)] from S1 for all process nodes. Although these numerical values ​​have different dimensions, they are mathematically given the same "comparable value" by multiplying them by their respective empirical weighting coefficients (α, β, γ), that is, they are all converted to the unified scale of "pressure contribution value".

[0056] This represents the static pressure resulting from the backlog. Each workpiece waiting in the queue continuously contributes to the base pressure value.

[0057] This represents the dynamic and cumulative pressure caused by waiting times. It more sensitively reflects the problem of stagnant flow—even if the queue is not long, an abnormal increase in workpiece waiting time indicates the presence of hidden obstacles.

[0058] (Regarding footwear production) This represents a specific pressure arising from imbalances in order. It successfully quantifies the qualitative problem of "mismatch" in the production system into a calculable quantitative indicator. The weight γ is typically set to a large value to ensure the system is sufficiently sensitive to the matching problem.

[0059] Adding the three pressure contribution values ​​above yields the comprehensive pressure value Pi(t) of process node i at time t. This value is dimensionless; its absolute magnitude is not important, but rather its relative level to the pressure values ​​of other process nodes. This value accurately characterizes the magnitude of the "resistance" that node poses to the smoothness of the entire production line at the current moment. Providing a clear and actionable basis for intelligent decision-making, the output of step S2 is a "real-time pressure spectrum" or "pressure map" of the entire production line. On this map, each process node corresponds to a specific pressure value. The profound significance of this pressure value Pi(t) lies in:

[0060] It enables the quantification and comparison of bottlenecks: the system can instantly identify the "maximum pressure node" of the entire line—that is, the most urgent bottleneck point—by comparing all Pi(t) without any ambiguity.

[0061] It reveals the nature of the bottleneck: by analyzing the composition of Pi(t), the system can infer the main cause of the bottleneck. For example, if a node Pi(t) is high and is mainly contributed by Ci(t), then the system knows that this is a "color mismatch" type bottleneck, and can trigger targeted scheduling instructions (such as prioritizing material pairing) instead of simply ordering the upstream to slow down.

[0062] Therefore, step S2 abstracts the complex production line status into a concise, decision-driving numerical signal. Like a sophisticated dashboard, it transforms the health status of the production system into a clear pointer reading, enabling subsequent intelligent scheduling to move beyond fuzzy experience and instead rely on precise, quantifiable data-driven logic.

[0063] Step S3: Identify the node with the greatest pressure.

[0064] This step is the decision-making linchpin of this method, and its role is to locate key issues from the macro-level situation. It does not perform calculations or measurements itself, but rather executes a critical, strategic judgment.

[0065] The output of step S2 is a dataset containing the real-time comprehensive pressure value Pi(t) for each process node i on the production line, i.e., {P1(t), P2(t), P3(t), ..., Pn(t)}. This "pressure map" comprehensively describes the load state of the entire system at time t. However, the energy for managing resources and solving problems is limited, and spreading efforts across the board often yields less result. Therefore, the system must have a mechanism to quickly, accurately, and unambiguously identify the most critical focus—the point that poses the greatest threat to the overall production flow—from this comprehensive dataset. Identifying this point is a prerequisite for taking any targeted scheduling actions.

[0066] Step S3 is logically an extreme value search algorithm. Its execution process is straightforward and synchronized with the calculation cycle of S2.

[0067] The algorithm initializes a temporary variable P_max and sets its value to a minimum value (such as 0 or negative infinity), while also initializing a variable BottleneckNode to record the identifier of the node with the maximum pressure.

[0068] The algorithm begins by traversing the set of pressure values ​​{P1(t), P2(t), ..., Pn(t)} generated by S2. For each pressure value Pi(t) in the set, the algorithm compares it with the currently recorded P_max.

[0069] If Pi(t) is greater than the currently recorded P_max, the algorithm performs two actions: first, it updates the value of P_max to this larger Pi(t); second, it updates the value of BottleneckNode to the number of the corresponding node i (e.g., "Node 3: Left thread of the sewing machine"). This step ensures that the system always records the maximum value and its identity among the currently traversed nodes.

[0070] Traversal completion and confirmation: After the algorithm completes the traversal of the entire pressure set, the variable P_max stores the maximum pressure value among all nodes in the entire line, while BottleneckNode stores the identity of the node to which this maximum pressure value belongs.

[0071] This process is computationally very efficient. Its core is a simple loop comparison that ensures that even if there are hundreds of nodes on the production line, identification can be completed within milliseconds.

[0072] The output of step S3 is identified as the node P_max, serving as the "compass" for the entire scheduling system. This identification result is of great significance:

[0073] This makes scheduling actions targeted: all subsequent control commands (S4 steps) generated by the system will revolve around this P_max node. Whether it's commanding its upstream to slow down or urging its downstream to accelerate, all the "thrust" is applied to alleviate this point of maximum pressure.

[0074] Because this identification process is repeated periodically, the system can dynamically track the shifting of bottlenecks on the production line. A bottleneck might be on the sewing line one minute, and then shift to the molding line the next minute due to improved flow. This dynamic tracking capability is something traditional static scheduling methods completely lack.

[0075] Step S4: Generate intelligent scheduling and control instructions.

[0076] This step is the "decision execution" stage of this method, which transforms the identification results of S3 into a series of specific and executable action instructions.

[0077] The core logic of the S4 step stems from a clever idea: instead of directly addressing the bottleneck itself, it manages its inputs and outputs. Its decision-making process is as follows:

[0078] Applying a "throttling" thrust upstream (first instruction): The system immediately sends a "decelerate" or "pause" instruction to the directly upstream process of node P_max. The underlying logic is: since the bottleneck is there, continuing to feed workpieces to it will only exacerbate the congestion. By temporarily halting upstream material feeding, time is gained for the bottleneck node to process the backlog of workpieces. The magnitude of the deceleration can be proportional to (P_max - P_avg) / P_avg (i.e., the relative proportion of pressure exceeding the average level), achieving precise control of the force.

[0079] Applying a "dredging" thrust downstream (second instruction): Simultaneously, the system sends "accelerate" or "ready to receive" prompts to all downstream processes of the P_max node. The logic is: to clear the space downstream of the bottleneck node as quickly as possible, creating an unobstructed channel for the flow of completed workpieces, thereby reducing its output resistance.

[0080] Enhanced Decision Making (Third Instruction) for Footwear Production: This is a crucial branch of innovative decision-making. The system checks the color mismatch Ci(t) of the P_max node. If Ci(t) exceeds a preset threshold (e.g., 0.2), the system determines that the congestion at this node is primarily caused by "mismatches," rather than absolute insufficient capacity. At this point, the system generates and issues a third instruction. This instruction no longer merely adjusts the takt time, but rather the production sequence. The instruction is sent to all upstream processes of the P_max node (up to the initial feeding station), requiring them to prioritize processing workpieces that can be paired with "lone shoes" in the P_max node queue. For example, the screen might display: "Prioritize: Color code - 24AW - BLUE, Size - 42, Foot type - R." This is a form of "precise dredging," addressing the pairing problem at its source.

[0081] The output of step S4 is a set of specific control commands sent to specific process nodes (upstream, downstream, or even the global material feeding end). These commands are no longer internal data, but rather specific action guidelines that will be applied to the physical production environment.

[0082] Step S5: Command execution and visualization.

[0083] Step S5 is the final interface for human interaction with this method, representing the "final step" in translating data-driven decisions into action in the physical world. Its core mission is to transform the digital instructions generated in S4 and stored on the server into intuitive signals that operators on the production line can understand instantly and execute precisely without complex thinking.

[0084] The S4 step generates a series of precise scheduling instructions, such as "command node A to reduce the beat rate by 10%" or "command node B to prioritize the blue 42-yard right foot." However, these instructions are machine-readable logical commands and cannot be directly transmitted to the operator. In traditional systems, such instructions might need to be relayed by the team leader via shouting or telephone, which is inefficient and error-prone. Therefore, an efficient and unambiguous "translation" and "presentation" mechanism is essential to deliver the right instructions to the right person at the right time and in the right way.

[0085] Step S5 achieves this goal through "adaptive metronomes" (usually electronic displays) deployed at each workstation. The execution process is as follows:

[0086] The workstation display receives specific instruction packets sent to this workstation via the network. The instruction packet contains the instruction type (acceleration / deceleration / normal / priority), intensity parameters, and possible specific content (such as color codes).

[0087] The screen background color changes to provide the most intuitive status indication. Green background: indicates "standard cycle time," everything is normal. Yellow background with an upward arrow: indicates an "accelerate" command. Upon seeing this, the operator will strive to slightly increase the operating speed while maintaining quality. Red background with a downward arrow or pause icon: indicates a "decelerate" or "pause" command. The operator will slow down the pace or temporarily stop picking up materials from upstream.

[0088] At the center of the screen is a core area, not a static number, but a dynamic graphical metronome. For example, an aperture contracts from the outside in at specific intervals, or a progress bar fills in cycles. The length of this cycle is determined by the system's calculated ideal beat. The operator's goal is to complete the current task when the aperture contracts to the center (or the progress bar fills). This directly transforms the abstract concept of "beat" into a visual, followable rhythm.

[0089] When a third instruction (priority processing) is received, clear text will be displayed prominently on the screen, such as: "Priority Instruction: Process [24AW-BLUE-42-R]". This provides irrefutable, specific task information, ensuring that the operator processes the correct workpiece.

[0090] The output of step S5 is the operator's accurate and timely action. Only when the operator adjusts their work rhythm and content according to the visual prompts on the screen is the closed-loop control of the entire intelligent scheduling system truly completed.

[0091] Steps S6 and S7: Global feeding cycle control.

[0092] Steps S6 and S7 together constitute the "master valve" controller of the entire production system. They are based on a higher dimension than inter-process scheduling—that is, the input end of the production line. Their core responsibility is to intelligently adjust the initial feeding speed according to the health status of the final output end, thereby preventing the production system from "clogging" due to overfeeding or "starving" due to insufficient feeding at the global level.

[0093] Background: The need for global balance beyond local optimization

[0094] Steps S3-S5 address flow balancing within the production line by eliminating local bottlenecks through "fine-tuning" the takt time between processes. However, these internal optimizations may fail if the source of the production line—material input—is not controlled. Consider two extreme cases:

[0095] Overfeeding: Even with the most intelligent internal scheduling, if the feeding speed continuously exceeds the production line's maximum capacity, the entire system will eventually become like a clogged pipe, causing backlogs in all processes, a sharp increase in work-in-process inventory, and unnecessary oscillations in the internal scheduling system, failing to fundamentally solve the problem.

[0096] Insufficient material input: If the material input speed is too slow, although the production line will be very smooth (the pressure value is very low), the final output efficiency will be low, the equipment and manpower will be idle, and the production target cannot be achieved.

[0097] Therefore, a feedback mechanism based on the final output is necessary to macroscopically regulate the material input, the "master switch." Its operational process is as follows:

[0098] S6: Monitor the final export status

[0099] The system continuously monitors the pressure value at the final stage of the production process, typically the packaging, quality inspection, or warehousing stations, and we denot this pressure value as P_end(t). This P_end(t) is a key indicator of whether the final output of the production line is running smoothly. A low P_end(t) or even zero indicates that the finished product is quickly packaged and transported, and the downstream flow is smooth. However, it could also mean that the production line is underutilized, with insufficient upstream material supply, and an impending break in flow. A consistently high P_end(t) indicates that the finished product outflow rate is slower than the production rate, resulting in a backlog at the final stage. The reasons could be insufficient storage space, inadequate handling capacity, or simply that the material input rate exceeds the actual maximum capacity of the entire line.

[0100] S7: Adjust the feeding cycle.

[0101] The system compares the value of P_end(t) with two preset thresholds (threshold_Low and threshold_High) and makes a decision accordingly.

[0102] If P_end(t) < threshold_Low: The final outlet pressure is too low, the production line faces the risk of "shutdown," and the capacity is not fully utilized. An "acceleration" command is issued to the feeding process. The command strength can be related to the degree to which the pressure is below the threshold, slightly increasing the feeding cycle to "replenish ammunition" for the system.

[0103] If P_end(t) > threshold_High: This indicates a blockage at the final outlet, meaning the entire supply chain from material input to output is overwhelmed. Continuing to input materials will only lead to a vicious cycle of work-in-process inventory. Issue a "slow down" or "pause" instruction to the material input process. This is the most critical protective measure, like turning off a tap to prevent further congestion.

[0104] If threshold_Low <= P_end(t) <= threshold_High: the final outlet is in the healthy pressure range, and the outflow is smooth. The instruction feeding process feeds materials according to the standard target cycle time to maintain the current stable production status.

[0105] Ultimately, this achieves global system stability and maximizes system productivity.

[0106] To evaluate the actual effectiveness of the scheduling method of this invention, we conducted a four-week comparative test on a medium-sized molding production line. The main processes of this production line include: pre-production preparation, sewing line, molding line, and quality inspection and packaging. The test aimed to quantitatively compare the differences between the new method and the traditional push scheduling based on fixed-day plans (hereinafter referred to as the "traditional method") in key indicators such as production efficiency, work-in-process inventory, and on-time order delivery rate.

[0107] Test production line: The same physical production line, with a relatively stable number of employees and skill levels.

[0108] Product: Two types of athletic shoes (Model A and Model B) were produced during the testing period, each with three color options. Daily order volume fluctuated to simulate real market demand.

[0109] Test period: 4 weeks in total. The first two weeks will use traditional fixed-cycle scheduling (as a baseline), and the last two weeks will switch to intelligent scheduling methods.

[0110] Data Acquisition: Use the same MES system to collect basic production data to ensure data source consistency.

[0111] Traditional method setup: Fixed production rhythm is set based on historical experience, and production is carried out sequentially after daily production plan is scheduled. Team leaders handle abnormalities based on experience.

[0112] Intelligent method settings: α (queue length) = 1.0, β (waiting time) = 0.1, γ (color mismatch) = 2.0. Pressure calculation cycle: 10 seconds. Feeding end pressure control thresholds: threshold_Low = 0.5, threshold_High = 3.0.

[0113] The test data is summarized in the following comparison table:

[0114]

[0115] Test Result Analysis:

[0116] Efficiency Improvement Analysis: The average daily output increased by 10.7%, primarily due to the intelligent methods breaking the limitations of traditional fixed-cycle production. Through real-time pressure sensing, it dynamically eliminates minor delays and blockages between processes, making the production flow more continuous and thus completing more work in the same amount of time. A significant 23% reduction in production cycle time is a direct reflection of this efficiency improvement.

[0117] Liquidity optimization analysis: One of the most significant benefits of this invention is the 32.1% reduction in work-in-process inventory. This is directly attributed to the global material feeding control of the S600 / S700 and the internal cycle time balancing of the S300-S500. The system acts like an intelligent traffic control system, avoiding congestion and allowing materials to pass quickly through the production line, reducing capital occupation and space pressure.

[0118] Analysis of solutions to unique problems in the footwear industry: The 72.6% reduction in "isolated shoe" inventory is highly convincing. This clearly demonstrates the effectiveness of introducing "color mismatch" (γ factor). The system no longer passively waits for problems to occur, but proactively issues warnings and allocates resources (third-party instructions) to prevent mismatch issues at the source, solving pain points that traditional methods cannot address.

[0119] Management effectiveness analysis: The sharp decrease in the number of interventions by team leaders indicates that the system has successfully automated and intelligently implemented most routine scheduling decisions. This frees managers from the heavy burden of "firefighting" tasks, allowing them to focus on higher-value work such as process improvement and personnel training.

[0120] In summary, under the industrial environment set up in this test, the "Dynamic Pressure Flow" intelligent scheduling method significantly outperforms the traditional fixed-cycle scheduling method in all key performance indicators. This method not only improves production efficiency and flow through data-driven dynamic optimization, but more importantly, its innovative "color mismatch" mechanism precisely solves the core problem of "color matching" in discrete footwear manufacturing.

[0121] Test results show that the present invention has industrial feasibility, significant economic benefits (increased output and reduced inventory), and unique industry relevance. It is a clever and efficient innovative solution for production scheduling and control, and has high value for promotion and application.

[0122] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for intelligent scheduling and control of footwear production cycle time based on MES data-driven approach, characterized in that, Includes the following steps: Step S1: Collect production status data of each process node on the production line in real time. The production status data includes at least the length of the work-in-process queue in front of each process node and the waiting time of each workpiece in the queue. Step S2: Based on the production status data, dynamically calculate the real-time pressure value of each process node; Step S3: Identify the process node with the highest real-time pressure value across the entire line and record it as the maximum pressure node; Step S4: Based on the location and pressure value of the maximum pressure node, generate a scheduling control instruction, wherein the scheduling control instruction includes at least: issuing a first instruction to the direct upstream process node of the maximum pressure node, the first instruction being used to reduce the production cycle time of the upstream process node; and issuing a second instruction to the downstream process node of the maximum pressure node, the second instruction being used to accelerate the production cycle time of the downstream process node. Step S5: Send the scheduling control command to the terminal display device of the corresponding process node for visual prompts to guide the operator to execute it.

2. The intelligent scheduling and control method for footwear production cycle based on MES data-driven approach according to claim 1, characterized in that, In step S2, the real-time pressure value Pi(t) is calculated using the following formula: ;in, Pi(t) is the pressure value of process node i at time t, Li(t) is the queue length of process node i at time t, Ti(t) is the average waiting time of the workpiece in the queue of process node i at time t, and α and β are preset weighting coefficients.

3. The intelligent scheduling and control method for footwear production cycle based on MES data-driven methods according to claim 1 or 2, characterized in that, In step S4, the rate of decrease in the cycle time of the first instruction is positively correlated with the magnitude by which the pressure value of the maximum pressure node exceeds the average pressure value of the entire line; and / or, the rate of increase in the cycle time of the second instruction is positively correlated with the magnitude by which the pressure value of the downstream process node of the maximum pressure node is lower than the average pressure value of the entire line.

4. The intelligent scheduling and control method for footwear production cycle based on MES data-driven approach according to claim 1, characterized in that, It also includes the control steps for the feeding cycle of the production line: Step S6: Monitor the pressure value of the final process node at the end of the production line in real time; Step S7: Dynamically adjust the feeding cycle at the beginning of the production line according to the pressure value of the final process node; when the pressure value of the final process node is lower than the first threshold, increase the feeding cycle; when the pressure value of the final process node is higher than the second threshold, decrease or suspend the feeding cycle.

5. The intelligent scheduling and control method for footwear production cycle time based on MES data-driven method according to claim 1, characterized in that, The visual prompts mentioned in step S5 include: distinguishing different instruction states by changing the background color of the terminal display device, wherein: green background represents normal rhythm; yellow background represents acceleration instruction; and red background represents deceleration or pause instruction.

6. The intelligent scheduling and control method for footwear production cycle time based on MES data-driven method according to claim 5, characterized in that, The visualization prompts also include dynamically displaying the theoretical operating rhythm of the current process on the terminal display device in the form of a metronome or countdown timer, in order to guide the operator's operating rhythm.

7. The intelligent scheduling and control method for footwear production cycle based on MES data-driven method according to claim 2, characterized in that, In step S1, the production status data also includes color code and size information determined based on the unique identification information of each workpiece; in step S2, the formula for calculating the real-time pressure value Pi(t) is: ;in, Ci(t) is the color mismatch degree of process node i at time t, which is the ratio of the number of workpieces in the queue that cannot be matched based on the color code and size information to the total length of the queue Li(t); γ is a preset weighting coefficient.

8. The intelligent scheduling and control method for footwear production cycle time based on MES data-driven method according to claim 7, characterized in that, When the system identifies a certain process node as the node with the greatest pressure and its color mismatch Ci(t) exceeds the third threshold, the scheduling control instruction generated in step S4 further includes: issuing a third instruction to the upstream process node of the node. The third instruction is used to optimize the feeding or processing sequence and prioritize the processing of workpieces that can be paired with workpieces that cannot be paired in the current queue.

9. A footwear production cycle intelligent scheduling and control system for implementing the method of any one of claims 1 to 8, characterized in that, include: The sensing module is configured to collect production status data of each process node in real time. The decision module is configured to calculate the real-time pressure value of each node, identify the node with the highest pressure, and generate scheduling control instructions; the execution module includes terminal display devices set at each process node, configured to receive and visualize the scheduling control instructions.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the steps of the method as described in any one of claims 1 to 8.

Images