Automated production line manufacturing execution system and production scheduling method for motor vehicle lamps

Abstract

This invention relates to the field of automated manufacturing and production scheduling technology, specifically to a manufacturing execution system and scheduling method for automated production lines of automotive lighting systems. It includes edge data acquisition, spatiotemporal flow tensor calculation, gradient field scheduling, and underlying execution modules. The system determines the physicochemical decay state of semi-finished automotive lighting products by acquiring their physical environment parameters and residence time. Its core is to calculate the dynamic viscosity tensor based on the aforementioned parameters and combine it with a gravitational potential well representing the scheduling priority of workstations to construct a continuous gradient field of the production line topology. The system extracts the steepest descent path based on this gradient field and adaptively generates routing and diversion instructions. This invention achieves a transformation from rigid, static time-window scheduling to highly dynamic and adaptive precise routing, effectively preventing high-risk materials from exceeding their physicochemical failure threshold while waiting in queues.

Description

Technical Field

[0001] This invention relates to the field of automated manufacturing and production scheduling technology, specifically to an automated production line manufacturing execution system and production scheduling method for automotive lighting. Background Technology

[0002] In the current automated manufacturing lines for automotive lights, the semi-finished products undergo irreversible degradation of their physical and chemical properties during the continuous process from gluing to curing, due to the influence of ambient temperature, humidity, and residence time. To schedule such materials, existing manufacturing execution systems generally adopt a scheduling method based on static time windows and rely on delayed physical inspections in downstream processes. Although this approach can maintain the basic flow of production materials, it fails to incorporate the real-time physical and chemical degradation state of semi-finished products and the dynamic operating efficiency of workstations into the underlying control, resulting in severely delayed scheduling decisions and a lack of dynamic adaptability. This rigid scheduling method can cause the physical and chemical state of materials to exceed the physical and chemical failure threshold while waiting in the queue, and the continued processing of failed semi-finished products will ineffectively occupy the processing cycle of subsequent workstations. At the same time, it is impossible to achieve adaptive reverse adjustment of the entire line's processing cycle when local equipment malfunctions. Therefore, how to achieve precise control of production scheduling based on the real-time physicochemical decay state of semi-finished products and the efficiency of production line equipment has become an urgent technical problem to be solved. Summary of the Invention

[0003] To solve the above-mentioned technical problems, the present invention provides an automated production line manufacturing execution system and a production scheduling method for automotive lighting. Specifically, the technical solution of the present invention is as follows: The manufacturing execution system for an automated production line of automotive lighting includes: The edge data acquisition module is used to acquire in real time the current position node, physical environment parameters, and dwell time of the semi-finished automotive lamp in the continuous process at the current workstation corresponding to the current position node; wherein, the continuous process includes manufacturing links with irreversible physical and chemical property decay characteristics. The spatiotemporal rheological tensor calculation module is used to input physical environment parameters and residence time into a preset decay dynamics equation and map them into a dynamic viscosity tensor; among which, the dynamic viscosity tensor is used to characterize the physicochemical decay state of the semi-finished vehicle lamp. The gradient field scheduling module is used to map the equipment operating efficiency value of each workstation in the automated production line as a depth parameter to a gravitational potential well that represents the scheduling priority of the workstation, and to construct the topological continuous gradient field corresponding to the automated production line based on the dynamic viscosity tensor and the gravitational potential well. The underlying execution module is used to extract the steepest descent path to the preset target workstation based on the continuous gradient field, starting from the current position node, and generate routing instructions and diversion instructions based on the steepest descent path and send them to the corresponding execution devices to control the semi-finished vehicle lights to enter the target workstation.

[0004] A production scheduling method for an automated production line of automotive lighting, applied to the manufacturing execution system for an automated production line of automotive lighting as described in claim 1, includes: Obtain the current position node, physical environment parameters, and dwell time of the semi-finished vehicle light product at the current workstation corresponding to the current position node in a continuous process. The physical environment parameters and residence time are input into the preset decay dynamics equation and mapped to a dynamic viscosity tensor. The equipment operating efficiency value of each workstation in the automated production line is used as a depth parameter and mapped to a gravitational potential well that represents the scheduling priority of the workstation. Based on the dynamic viscosity tensor and the gravitational potential well, a topological continuous gradient field corresponding to the automated production line is constructed. Based on the continuous gradient field, starting from the current position node, extract the steepest descent path to the preset target workstation; Routing and diversion instructions are generated based on the steepest descent path to control the entry of semi-finished vehicle lights into the target workstation.

[0005] Optionally, physical environment parameters and residence time are input into a preset decay kinetic equation and mapped to a dynamic viscosity tensor, including: Extract pre-defined order data from the manufacturing execution system; Initialize the rheological tensor baseline of the semi-finished automotive headlights based on order data from the Manufacturing Execution System; Input the physical environment parameters and residence time into the preset decay kinetic equation, calculate the product of the deviation coefficient of the physical environment parameters relative to the reference process parameters and the residence time, and multiply it by the preset decay weight matrix to obtain the decay evolution value of the matrix dimension. The rheological tensor baseline and the decay evolution value are fused by matrix addition to generate a dynamic viscosity tensor.

[0006] Optionally, the equipment operating efficiency value of each workstation in the automated production line is used as a depth parameter and mapped to a gravitational potential well representing the workstation scheduling priority. A topological continuous gradient field corresponding to the automated production line is constructed based on the dynamic viscosity tensor and the gravitational potential well, including: The overall equipment efficiency feedback value of each workstation in the automated production line is obtained through a programmable logic controller and used as the equipment operating efficiency value. The reciprocal of the overall equipment efficiency feedback value is used to determine the potential depth of the gravitational potential well. The potential well depth is multiplied by the dynamic viscosity tensor to generate a viscosity gradient field. The viscosity gradient field is output as a continuous gradient field.

[0007] Optionally, before extracting the steepest descent path to the preset target workstation based on the continuous gradient field and starting from the current position node, the method further includes: Calculate the matrix norm of the dynamic viscosity tensor and compare it with the preset critical solidification threshold; If the matrix norm is greater than or equal to the critical solidification threshold, a scrap rejection instruction is generated and sent to the corresponding robotic arm controller to control the sorting robotic arm configured in the automated production line to remove the semi-finished car light product from the automated production line. If the matrix norm is less than the critical solidification threshold, then the step of extracting the steepest descent path to the preset target workstation based on the continuous gradient field and starting from the current position node is executed.

[0008] Optionally, it also includes: Real-time monitoring of the fault status of each workstation on the automated production line; If a fault is detected at any workstation in the automated production line, the workstation with the fault is designated as an abnormal workstation. Based on the gradient difference from the abnormal workstation to its upstream node in the continuous gradient field, the reverse transmission pressure value is obtained by multiplying the gradient difference with the preset distance attenuation coefficient. The reverse transmission pressure value is compared with the preset pressure threshold; If the reverse transmission pressure value is greater than or equal to the pressure threshold, a speed reduction command is generated, and the processing equipment corresponding to the station upstream of the abnormal station is controlled to reduce the cycle time parameter. If the reverse transmission pressure value is less than the pressure threshold, the current cycle parameters of the processing equipment are maintained.

[0009] Optionally, the current position node and physical environment parameters of the semi-finished vehicle light product in a continuous process are obtained, as well as the dwell time of the semi-finished vehicle light product at the current workstation corresponding to the current position node, including: Get the current location node and the dwell time; The temperature and humidity sensor array is controlled by the edge computing gateway to collect data according to a preset sampling period; wherein the preset sampling period is less than or equal to fifty milliseconds. Based on the data acquisition results, ambient temperature and humidity data were extracted. Ambient temperature and humidity data are used as physical environmental parameters.

[0010] Optionally, routing instructions and diversion instructions are generated based on the steepest descent path to control the entry of the semi-finished vehicle headlights into the target workstation, including: Determine the target direction based on the steepest descent path; The target direction is encapsulated into routing and routing commands; The routing command is sent to the automated guided vehicle (AGV) scheduling interface configured on the automated production line to control the AGVs configured on the automated production line to move to the target direction; The diversion command is sent to the diversion baffle actuator configured in the automated production line to control the conveyor belt diversion baffle configured in the automated production line to switch to the target direction.

[0011] Compared with the prior art, the present invention has the following beneficial effects: 1. This invention obtains the physical environment parameters and residence time of semi-finished automotive lights in continuous processes, inputs them into a preset decay dynamics equation and maps them to a dynamic viscosity tensor. At the same time, it maps the equipment operating efficiency values ​​of each workstation in the automated production line to a gravitational potential well that characterizes the scheduling priority of the workstation, constructs a continuous gradient field and extracts the steepest descent path to the target workstation. This mechanism breaks the traditional rigid static time window scheduling method, directly incorporates the real-time irreversible physical and chemical decay state of semi-finished automotive lights and the dynamic operating efficiency of the workstation equipment into the underlying control, realizes highly dynamic and adaptive precise routing and diversion, and effectively avoids high-risk materials from exceeding the physical and chemical failure threshold while waiting in the queue. 2. This invention calculates the matrix norm of the dynamic viscosity tensor and compares it with a preset critical solidification threshold. If the matrix norm is greater than or equal to the critical solidification threshold, a scrap rejection command is directly generated, controlling the sorting robotic arm configured in the automated production line to remove the semi-finished automotive lamp. This mechanism overcomes the shortcomings of existing technologies that rely on delayed physical detection in later processes. It can perform dynamic prediction and active isolation of semi-finished products that have failed in their physical and chemical state before the extraction path, avoiding the continued ineffective transfer and processing of failed semi-finished products, thereby reducing the ineffective processing load of subsequent stations. 3. This invention monitors the fault status of each workstation in an automated production line in real time. When a fault is detected at any workstation, the reverse transmission pressure value is calculated based on the gradient difference from the abnormal workstation to its upstream node in a continuous gradient field and a preset distance attenuation coefficient. When the reverse transmission pressure value is greater than or equal to a preset pressure threshold, a deceleration command is generated and the processing equipment located upstream of the abnormal workstation is controlled to reduce its cycle time parameter. This mechanism solves the problem of lack of dynamic adjustment capability for the entire line when local equipment malfunctions, transforms local faults into quantifiable reverse control signals, and realizes dynamic contraction control of the output cycle time of upstream equipment, effectively preventing disordered accumulation and obstructed flow of materials in the production line. 4. In the process of generating the dynamic viscosity tensor, this invention extracts preset manufacturing execution system order data to initialize the rheological tensor baseline of the automotive lamp semi-finished product, calculates the decay evolution value by multiplying the physical environment parameters and residence time, and fuses the rheological tensor baseline and the decay evolution value by matrix addition. This mechanism explicitly introduces the process differences of the order dimension into the underlying data structure modeling, avoiding the problem of ignoring the initial process differences and applying the same decay starting point to different products, thereby providing more accurate decay state characterization data for subsequent gradient field production scheduling. Attached Figure Description

[0012] The present invention will be further explained below with reference to the accompanying drawings and embodiments: Figure 1 This is a structural diagram of the system of the present invention; Figure 2 This is a flowchart of the method of the present invention. Detailed Implementation

[0013] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments. Example 1: Please see Figure 1 The automated manufacturing execution system for automotive lighting production lines includes:

[0014] The edge data acquisition module is used to acquire in real time the current position node, physical environment parameters, and dwell time of the semi-finished automotive lamp in the continuous process at the current workstation corresponding to the current position node; wherein, the continuous process includes manufacturing links with irreversible physical and chemical property decay characteristics. The spatiotemporal rheological tensor calculation module is used to input physical environment parameters and residence time into a preset decay dynamics equation and map them into a dynamic viscosity tensor; among which, the dynamic viscosity tensor is used to characterize the physicochemical decay state of the semi-finished vehicle lamp. The gradient field scheduling module is used to map the equipment operating efficiency value of each workstation in the automated production line as a depth parameter to a virtual gravitational potential field depth that characterizes the scheduling priority of the workstation, hereinafter referred to as the potential well, and to construct a continuous gradient field of the automated production line topology based on the dynamic viscosity tensor and the gravitational potential well. The underlying execution module is used to extract the steepest descent path to the preset target workstation based on the continuous gradient field, starting from the current position node, and generate routing instructions and diversion instructions based on the steepest descent path and send them to the corresponding execution devices to control the semi-finished vehicle lights to enter the target workstation.

[0015] This embodiment provides an implementation mechanism for an automated production line manufacturing execution system for automotive lighting. Specifically, the system is deployed in a mixed manufacturing line that includes an injection molding area, an adhesive coating area, an ultraviolet curing oven, an assembly area, and an inspection area. The semi-finished lamp housings after adhesive coating undergo irreversible degradation of their physical and chemical properties before entering the UV curing process. The system uses an edge computing gateway, a temperature and humidity sensor array, a programmable logic controller, an automated guided vehicle (AGV) scheduling interface, and a conveyor belt diversion baffle actuator as hardware carriers to perform real-time scheduling of the flow of semi-finished products in continuous processes. The details are as follows: The edge data acquisition module obtains the current position node of each lamp housing semi-finished product; the current position node can be any one of the discrete nodes such as the glue coating robot exit, the buffer conveyor section, the UV curing oven entrance waiting position; at the same time, the edge data acquisition module also reads the ambient temperature, ambient humidity of the area where the node is located, and the time that the semi-finished product has stayed at the current workstation; for ease of understanding, each semi-finished product can be regarded as a flow unit with state parameters; for example, a certain lamp housing semi-finished product A is currently located in the first main curing oven of the buffer section after glue coating, the ambient temperature is 31 degrees Celsius, the ambient humidity is 68%, and the residence time is 42 seconds; After receiving the above data, the spatiotemporal rheological tensor calculation module inputs the physical environment parameters and residence time into a preset decay kinetic equation. This equation establishes a physicochemical decay increment model based on environmental temperature deviation, environmental humidity deviation, and residence time, and superimposes the base state to generate a dynamic viscosity tensor. The tensor here does not require a complex high-order expression; in engineering implementation, a two-dimensional matrix can be used to represent its decay state in multiple physicochemical dimensions. For example, a 2×2 matrix can be used to represent the two directions of the semi-finished product: the tendency for surface crusting and the tendency for internal fluidity to decrease. The initial baseline is set as a second-order tensor consisting of elements 0.20 and 0.05 forming the first row and elements 0.04 and 0.18 forming the second row. If the decay evolution value obtained based on the current temperature, humidity, and residence time is [[0.16,0.02],[0.01,0.14]], then the dynamic viscosity tensor obtained after fusion is [[0.36,0.07],[0.05,0.32]]. This result indicates that the semi-finished product has already exhibited physicochemical decay exceeding the preset decay threshold compared to when the adhesive was first applied, and the priority of subsequent scheduling should be increased. The gradient field scheduling module further maps the equipment operating efficiency values ​​of each workstation on the production line to gravitational potential wells. The equipment operating efficiency values ​​can be obtained from the overall equipment efficiency fed back by the programmable logic controller. For example, the equipment operating efficiency value of UV curing oven B workstation is 0.92, and the equipment operating efficiency value of the standby UV curing oven C workstation is 0.80. Their reciprocals can be approximated as potential well depths of 1.09 and 1.25, respectively. When the potential well depth value is greater than the set depth threshold, it indicates that the workstation currently has the scheduling priority to accept semi-finished products with dynamic viscosity tensors exceeding the preset value. The dynamic viscosity tensor corresponding to the semi-finished product is fused with the potential well depth of each candidate workstation to construct a continuous gradient field on the production line topology. In engineering implementation, the connection relationship between each node can be represented as a topology graph, and then a gradient value can be assigned to each edge. The steeper the gradient value, the more beneficial it is to reduce the risk of failure when traveling from the current node to the target workstation. The underlying execution module extracts the steepest descent path to the target workstation from the current position node based on a continuous gradient field. For ease of understanding, it is assumed that the first main curing oven at the current position node can lead to two directions: path P1 leads to UV curing oven B, with a comprehensive gradient descent value of 0.18; path P2 leads to the backup UV curing oven C, with a comprehensive gradient descent value of 0.31. In this case, the system selects P2 as the steepest descent path. The underlying execution module generates routing instructions and diversion instructions corresponding to this path. The former is sent to the automated guided vehicle scheduling interface to control the AGV carrying the semi-finished product to drive to the entrance of workstation C; the latter is sent to the conveyor belt diversion baffle actuator to switch the baffle of the buffer conveyor section to branch C. Regarding the anomaly handling mechanism, if a sensor at a certain node goes offline for a short period of time, the edge data acquisition module can call up the valid data from the previous sampling period and add a timestamp for verification; if it fails to recover for two consecutive sampling periods, the semi-finished product is marked as incomplete, and the system will not execute high-risk cross-regional flow temporarily, but will instead prioritize directing it to the nearest manual review buffer; if the PLC feedback of a candidate workstation is abnormal, the gradient field scheduling module sets the potential well depth of that workstation to an invalid high value to exclude it from the candidate set; if all candidate workstations are in an unavailable state, the underlying execution module outputs a delay instruction, causing the semi-finished product to enter the controlled buffer and continuously recalculate the gradient field until a feasible path appears; For example, in a headlight assembly production line, a batch of LED headlight housings was coated with adhesive at 18:03:12. The 17th semi-finished product was queued in the buffer section by an automated guided vehicle. Within a 50-millisecond sampling period, the system detected that the temperature in the area where the semi-finished product was located was rising and the residence time was approaching the upper limit of the process. It then calculated that the rate of change of its dynamic viscosity tensor exceeded the preset tensor amplification threshold. At this time, although the first main curing oven was close to the main curing oven, its operating efficiency decreased due to the changeover, resulting in a decrease in its potential well depth. The first backup curing oven had high operating efficiency, and its potential well depth was significantly greater than that of the former. After multiplying and summing the potential well depths of the two with the current dynamic viscosity tensor of the semi-finished product, the first backup curing oven generated a larger comprehensive gradient descent value. Based on this, the system diverted the 17th semi-finished product to station C to prevent it from failing due to surface skinning before entering the curing stage. By directly incorporating the physical and chemical decay state of semi-finished automotive lights into the underlying data structure of the scheduling, production scheduling decisions no longer rely solely on static time windows, but rather on continuously updated physical and chemical field changes, thereby achieving real-time control of hybrid manufacturing production lines with irreversible physical and chemical constraints.

[0016] Example 2: Please see Figure 2 A production scheduling method for automated production lines of automotive lighting, applied to the manufacturing execution system of automated production lines of automotive lighting as described in the embodiment, includes: Obtain the current position node, physical environment parameters, and dwell time of the semi-finished vehicle light product at the current workstation corresponding to the current position node in a continuous process. The physical environment parameters and residence time are input into the preset decay dynamics equation and mapped to a dynamic viscosity tensor. The equipment operating efficiency value of each workstation in the automated production line is used as a depth parameter and mapped to the depth of a virtual gravitational potential field that characterizes the scheduling priority of the workstation, hereinafter referred to as the potential well. A continuous gradient field of the automated production line topology is constructed based on the dynamic viscosity tensor and the gravitational potential well. Based on the continuous gradient field, starting from the current position node, extract the steepest descent path to the preset target workstation; Routing and diversion instructions are generated based on the steepest descent path to control the entry of semi-finished vehicle lights into the target workstation.

[0017] This embodiment provides a production scheduling step for an automated production line of automotive lighting. Specifically, this step follows the aforementioned hybrid manufacturing main line scenario for automotive lighting assemblies, with the semi-finished lamp housings awaiting curing after adhesive application as the key scheduling object. Millisecond-level scheduling is completed through a chain process of acquisition, mapping, field construction, path finding, and execution. The details are as follows: Obtain the node position, physical environment parameters, and residence time of the current semi-finished product; the position here does not only represent spatial coordinates, but also the process node identity in the production line topology; for example, even if they are both in a waiting state, the feasible paths and risk levels are different if they are located in the direct-connection buffer section after gluing and the pre-inspection position at the entrance of the curing oven, so they need to be modeled in a node manner; The physical environment parameters and residence time are input into the preset decay kinetic equation and mapped to a dynamic viscosity tensor. To demonstrate the data flow process, it is assumed that the temperature of a certain semi-finished product B at the sampling time is 30 degrees Celsius, the humidity is 65%, and the residence time is 55 seconds. The system retrieves the corresponding decay model from the database based on the glue formula used in this batch, and the tensor value of the semi-finished product is updated from [[0.30,0.03],[0.02,0.28]] in the previous moment to [[0.41,0.05],[0.03,0.39]], indicating that the semi-finished product has entered a high-risk stage. The equipment operating efficiency values ​​of each workstation are mapped as depth parameters to gravitational potential wells, and a continuous gradient field is constructed together with the dynamic viscosity tensor. Here, the production line can be abstracted as a network graph composed of several nodes and edges, and each candidate target workstation in the graph has a potential well depth. If the equipment operating efficiency values ​​of workstations M, N, and Q are 0.95, 0.70, and 0.88, respectively, then the depth parameters can be obtained as approximately 1.05, 1.43, and 1.14, respectively. After fusing the dynamic viscosity tensor of semi-finished product B with these depth parameters, the comprehensive descent values ​​from the current position to each workstation direction can be calculated as 0.22, 0.37, and 0.26, respectively. It can be seen that although the equipment efficiency of workstation N is low, under the current field construction rules, its deep potential well can output a larger potential field guidance value to match the high viscosity semi-finished product, thereby forming the target scheduling direction. Based on the continuous gradient field, the steepest descent path to the target workstation is extracted from the current node. This extraction process can be achieved by comparing adjacent nodes step by step. For example, if the current node is the first main curing oven, the next hops it can reach are N3 and N4. From N3, it can continue to reach M and Q, and from N4, it can continue to reach N. After comparing the cumulative descent values ​​step by step, the system selects the first main curing oven → N4 → N as the output path, without having to completely rearrange the entire production line. The system generates routing and diversion commands based on the steepest descent path to control the entry of semi-finished products into the target workstation. If the starting node is in the automated guided vehicle (AGV) transport area, the AGV routing command is generated first. If the starting node is in the conveyor belt branching area, the baffle switching command is generated simultaneously. If both exist simultaneously, the diversion command is issued first, and the AGV movement command is issued after the baffle is confirmed to be in place, in order to avoid mechanical conflicts. In the anomaly handling mechanism, if the sampled environmental parameters change abruptly, such as the humidity value jumping from 60% to 95% within a sampling period, the system will first perform anomaly verification; if it is determined to be sensor noise, the sliding window mean will be used instead; if it is determined to be a real environmental fluctuation, the new value will be used immediately for tensor update; if the steepest descent path includes an occupied conveyor belt section, the system will temporarily mark that section as impassable and re-execute pathfinding on the remaining available edges; if there are no available paths, a buffer dwell instruction will be output and the recalculation frequency of the semi-finished product will be increased. For example, in the aforementioned LED automotive lighting production line, at 19:15:06, the 23rd lamp housing is coated with adhesive and enters the buffer section from the robot exit. The system first obtains its node position as G1, ambient temperature 29 degrees Celsius, humidity 63%, and dwell time 12 seconds. As time progresses to 19:15:54, its tensor value reaches the preset high-risk tensor range. At this time, the queue at the main curing oven entrance increases, and the backup curing oven branch line is in a high-availability state. Based on this, the system extracts the steepest descent path to the backup curing oven from node G1, controls the diversion baffle to turn the branch line, and simultaneously schedules the nearest idle AGV to take over the transfer. By integrating real-time perception, physical and chemical mapping, workstation status and execution control into a single continuous process, the production scheduling and diversion of semi-finished automotive lights within the physical and chemical aging window can be achieved. The physical environment parameters and residence time are input into a preset decay kinetic equation, which is mapped to a dynamic viscosity tensor, including: Extract pre-defined order data from the manufacturing execution system; Initialize the rheological tensor baseline of the semi-finished automotive headlights based on order data from the Manufacturing Execution System; Input the physical environment parameters and residence time into the preset decay kinetic equation, calculate the product of the deviation coefficient of the physical environment parameters relative to the reference process parameters and the residence time, and multiply it by the preset decay weight matrix to obtain the decay evolution value of the matrix dimension. The rheological tensor baseline and the decay evolution value are fused by matrix addition to generate a dynamic viscosity tensor.

[0018] This embodiment provides a refined generation mechanism for dynamic viscosity tensors. Specifically, in the aforementioned scheduling process, although directly calculating the tensor based solely on real-time temperature, humidity, and residence time can reflect the current decay, under mixed-line production conditions with multiple inventory units, the adhesive formulations, lamp body structures, and curing requirements corresponding to different orders are not the same. Ignoring these initial process differences will result in different lamp housing models being assigned the same decay starting point, affecting the accuracy of subsequent scheduling. Therefore, this embodiment introduces order data to initialize the rheological tensor baseline, and then fuses it with the real-time decay evolution value. The details are as follows: The system first extracts order data from the manufacturing execution system. The order data includes at least the product model, adhesive type, adhesive application trajectory category, target curing curve, and the maximum allowable physicochemical failure window. For different orders, the system pre-establishes different rheological tensor baselines. For ease of explanation, assume that order O1 corresponds to a matrix headlight, using type A polyurethane adhesive, with an initial baseline of [[0.12,0.02],[0.01,0.10]]; order O2 corresponds to a common LED taillight, using type B silicone adhesive, with an initial baseline of [[0.08,0.01],[0.01,0.07]]; it can be seen that the initial physicochemical state of different orders differs immediately after adhesive application. The real-time acquired temperature, humidity, and residence time are input into a preset decay kinetic equation to calculate the decay evolution value. In engineering implementation, complex differential forms are not required in the specification; it can be understood as an incremental matrix formed by the combined effects of temperature, humidity, and time. Specifically, the calculation rule for the decay product is as follows: calculate the difference between the real-time ambient temperature and the baseline process temperature, and the difference between the real-time ambient humidity and the baseline process humidity, respectively, and sum them weighted according to the corresponding sensitivity coefficients to obtain the comprehensive environmental deviation coefficient. This logic can be expressed by the following formula:

[0019] Among them, the comprehensive environmental deviation coefficient E is jointly determined by the deviation of the real-time ambient temperature T from the reference process temperature T0 and the deviation of the real-time ambient humidity H from the reference process humidity H0, and the temperature sensitivity coefficient. Humidity sensitivity coefficient These are used to describe the degree of influence of the corresponding environmental factors on the attenuation; among them, Its dimension is the reciprocal of temperature. The dimension of the coefficient is the reciprocal of humidity, so as to ensure that the comprehensive environmental deviation coefficient E is a dimensionless pure scalar. The comprehensive environmental deviation coefficient, the residence time scalar t, and the preset attenuation weight matrix W corresponding to the adhesive type are multiplied sequentially to generate a matrix-form attenuation evolution value. To ensure dimensionality consistency in the formula addition, the preset attenuation weight matrix W contains the dimension of the reciprocal of time, thus guaranteeing the attenuation evolution value. Physical dimensions and rheological tensor baseline M base The physical dimensions are kept consistent; and it is fused with the rheological tensor baseline; the fusion logic can be expressed by the following equation:

[0020] Among them, the generated dynamic viscosity tensor M dynamic By rheological tensor baseline M baseThe formula is obtained by superimposing the incremental terms formed by environmental deviation, residence time, and material weight; the dot in the formula represents scalar multiplication and scalar-matrix multiplication. For example, a semi-finished product of order O1 is left at 31 degrees Celsius and 68% humidity for 40 seconds. The product of the comprehensive environmental deviation coefficient and time is calculated. Combined with the weight matrix of type A adhesive, the decay evolution value is [[0.15,0.03],[0.02,0.13]]. Then, matrix addition with its baseline is performed to obtain the dynamic viscosity tensor as [[0.27,0.05],[0.03,0.23]]. If the environment is the same and the order is O2, the decay evolution value can vary depending on the material as [[0.10,0.02],[0.01,0.09]], which is then fused with the baseline to obtain... [[0.18,0.03],[0.02,0.16]]; This shows that the same field environment has different attenuation consequences for different product types; In the anomaly handling mechanism, if the manufacturing execution system does not return complete order data, such as missing the glue formula version number, the system uses the default baseline template of the product family and adds a low-confidence mark to the semi-finished product to increase conservatism in subsequent scheduling, such as prioritizing delivery to the nearest curing station; if a certain environmental parameter is missing when calculating the decay evolution value, only the existing parameters and residence time can be used for partial calculation, and the increment corresponding to the missing dimension is set to zero; if the number of missing dimensions exceeds the preset number, the fusion is paused and manual confirmation is initiated; if any element after matrix addition exceeds the preset upper limit, it is truncated by the saturation value to prevent abnormal data from causing the scheduling field calculation to overflow. For example, during the night shift production on the same automotive lighting production line, orders O1 and O2 were alternately put online. At 22:10, the 5th semi-finished product of batch O1 and the 9th semi-finished product of batch O2 both stayed in the post-coating buffer section for 38 seconds, with almost the same temperature and humidity. The system did not treat the two as having equal risk, but instead read their respective order data, initialized the baselines separately, and then superimposed the real-time decay evolution value. The results showed that the tensor increase corresponding to batch O1 was higher, so the scheduling module prioritized sending it to the main curing oven, while batch O2 could temporarily maintain its original queuing order. By explicitly incorporating process differences at the order level into the physicochemical state modeling, more accurate attenuation assessment can be achieved under multi-SKU mixed-line conditions, avoiding misjudging different models of headlights as the same type of scheduling object; The equipment operating efficiency values ​​of each workstation in the automated production line are used as depth parameters and mapped to the depth of a virtual gravitational potential field, hereinafter referred to as the potential well, representing the scheduling priority of the workstation. A continuous gradient field of the automated production line topology is constructed based on the dynamic viscosity tensor and the gravitational potential well, including: The overall equipment efficiency feedback value of each workstation in the automated production line is obtained through a programmable logic controller and used as the equipment operating efficiency value. The reciprocal of the overall equipment efficiency feedback value is used to determine the potential depth of the gravitational potential well. The potential well depth is multiplied by the dynamic viscosity tensor to generate a viscosity gradient field. The viscosity gradient field is output as a continuous gradient field.

[0021] This embodiment provides a refined construction mechanism for a continuous gradient field. Specifically, in the aforementioned process, if the target position is selected solely based on whether the workstation is idle, it is difficult to accurately characterize the dynamic changes in the actual processing capacity of the equipment. For example, although a curing oven is not shut down, its efficiency decreases after the model is changed. If semi-finished products with viscosity tensors exceeding the threshold are continued to be introduced into the workstation, it may cause the queuing time to exceed the upper limit allowed by the process. Therefore, this embodiment obtains the comprehensive efficiency feedback value of the equipment through a PLC, converts it into the gravitational potential well depth, and then combines it with the dynamic viscosity tensor to generate a viscosity gradient field. The details are as follows: The system reads the comprehensive efficiency feedback value of each workstation through the PLC. This feedback value is formed by combining indicators such as equipment utilization rate, performance efficiency, and yield rate. Assuming there are three candidate workstations at the same time: the feedback value of curing oven B is 0.90, the feedback value of curing oven C is 0.75, and the feedback value of compensation curing unit D is 0.60; the system takes their reciprocals respectively, and obtains potential well depths of approximately 1.11, 1.33, and 1.67. This reciprocal mapping method allows workstations with lower efficiency to exhibit deeper attraction positions under specific strategies, which can be used to undertake certain tasks that urgently need to be bypassed or used for backup compensation. Of course, in engineering implementation, limiting conditions can also be set, such as only allowing workstations in a processable state to participate in the mapping. The potential well depth is multiplied by the dynamic viscosity tensor of the semi-finished product to generate a viscosity gradient field; let the potential well depth at station i be D. i The potential well depth is obtained from the reciprocal of the overall equipment efficiency feedback value. Therefore, the element in the j-th row and k-th column of the matrix corresponding to workstation i in the viscosity gradient field can be represented as: For ease of explanation, if the current tensor of a certain semi-finished product is [[0.30,0.04],[0.02,0.28]], then for the main curing oven B, we can obtain [[0.333,0.0444],[0.0222,0.3108]]; for the standby curing oven C, we can obtain [[0.399,0.0532],[0.0266,0.3724]]; and for station D, we can obtain [[0.501,0.0668],[0.0334,0.4676]]; The system further projects these results onto the edges of each path in the production line topology map, forming a continuous gradient field that can be used for pathfinding. Since the subsequent stage of extracting the steepest descent path requires node comparison using an approximate scalar weight metric, the system will compress and map the output by summing the absolute values ​​of all elements of the aforementioned matrix-form viscosity gradient field into a single scalar value, namely the comprehensive gradient descent value S. i Its calculation logic is as follows:

[0022] Among them, S i D represents the overall gradient descent value leading to workstation i. i The value represents the potential well depth at workstation i, and the row index j and column index k correspond to the positions of elements in the viscosity gradient field matrix, respectively. This represents the summation of the absolute values ​​of all elements in the matrix. The dynamic viscosity tensor M represents dynamic The element in the j-th row and k-th column, symbol This indicates the absolute value operation; and the resulting comprehensive gradient descent value is assigned to the corresponding topological edge; if the current strategy emphasizes the prompt handling of high-risk items, a larger comprehensive gradient descent value represents a stronger guiding effect; if the emphasis is on load balancing, a load correction term can be superimposed on the comprehensive gradient descent value to avoid all high-risk items being centrally scheduled to the same workstation at the same time; In the anomaly handling mechanism, if the PLC feedback value jitters briefly, such as a large jump between 0.90 and 0.50 for three consecutive cycles, the system can first use de-jitter filtering to calculate the stable value and then perform reciprocal mapping. If the feedback value of a certain station is close to zero, directly taking the reciprocal will cause an abnormal amplification of the potential well depth. Therefore, the system sets a minimum effective feedback lower limit, such as 0.20. If it is lower than this value, it is directly determined that the station cannot participate in the field construction. If all stations are determined to be unable to participate, the system only retains the dynamic viscosity information of the semi-finished product and temporarily does not output the complete gradient field. Instead, it switches to a waiting or manual takeover mode. For example, during the night shift production of headlights, the overall efficiency of the main curing oven B, which is switching lamp types, drops from 0.93 to 0.76 according to the PLC feedback. At the same time, the standby compensation curing unit D is available but has a slow processing cycle. For a high-risk lamp housing whose tensor value has risen rapidly, the system finds that the scalar gradient in the direction of D is larger after potential well mapping, tensor multiplication and summation to convert it into a comprehensive gradient descent value. Therefore, the lamp housing is guided to station D in advance to avoid it queuing at station B for a long time and exceeding the physical and chemical failure window. By incorporating real-time changes in equipment operating capacity into continuous field modeling, we can achieve integrated scheduling of process risks and equipment status, rather than simply performing material handling based on static process routes. Based on the continuous gradient field, before extracting the steepest descent path to the preset target workstation, starting from the current position node, the process also includes: Calculate the matrix norm of the dynamic viscosity tensor and compare it with the preset critical solidification threshold; If the matrix norm is greater than or equal to the critical solidification threshold, a scrap rejection instruction is generated and sent to the corresponding robotic arm controller to control the sorting robotic arm configured in the automated production line to remove the semi-finished car light product from the automated production line. If the matrix norm is less than the critical solidification threshold, then the step of extracting the steepest descent path to the preset target workstation based on the continuous gradient field and starting from the current position node is executed.

[0023] This embodiment provides a mechanism for predicting scrap before path finding. Specifically, in the aforementioned process, even if an optimal path can be calculated, if the physical and chemical state of the semi-finished product itself has failed, continuing to transport and process it will only occupy subsequent production capacity and cause ineffective production. Therefore, before performing path extraction, this embodiment first determines whether the critical solidification state has been reached based on the matrix norm of the dynamic viscosity tensor, and removes it in advance when the threshold is reached. The details are as follows: The system first calculates the matrix norm of the dynamic viscosity tensor; for ease of engineering implementation, the square root of the sum of squares of the elements can be used as the criterion; the calculation formula is as follows: ;in, Let represent the Frobenius matrix norm of the dynamic viscosity tensor, and j and k represent the row and column indices of the matrix, respectively. This represents the element in the j-th row and k-th column of the dynamic viscosity tensor. This represents summing the squares of all elements in each row and column. This represents the square root operation; assuming the tensor of a certain semi-finished product is [[0.58,0.10],[0.08,0.55]], then its norm can be approximated as:

[0024] The calculated Frobenius norm is approximately 0.81; if the system's preset critical solidification threshold is 0.75, it indicates that the semi-finished product has reached or exceeded the failure threshold. In this case, the system no longer enters the steepest descent path extraction step, but directly generates a scrap rejection instruction and sends the instruction to the sorting robot arm controller; after receiving the instruction, the sorting robot arm grabs the semi-finished product when it passes the nearest executable rejection position and removes it from the main line, sending it to the isolation box or rework area; conversely, if the tensor of another semi-finished product in the same batch is [[0.32,0.05],[0.03,0.30]], its norm is about 0.44, which is lower than the threshold, then it continues to enter the subsequent gradient field path finding and diversion steps; Compared to not having this mechanism, relying solely on the airtightness test in the later stage to detect the failure of the parts has a significant lag. Especially when the glue has solidified after application but has not yet entered the assembly process, if the parts continue to be transported, assembled, covered, and locked along the line, it will excessively occupy the processing cycle of the subsequent stations. Therefore, it is necessary to implement dynamic prediction and rejection in the earlier stage. Regarding the anomaly handling mechanism, if the norm of a semi-finished product is close to the threshold, such as 0.749 or 0.751, the system can set a gray zone, for example, between 0.74 and 0.76 as the area to be confirmed. Within the area to be confirmed, the system can increase the sampling frequency and request adjacent vision stations to review the surface condition. If the review result supports failure, rejection is performed; if the review result is normal, release continues. If the sorting robot arm is currently occupied and cannot complete rejection immediately, the system temporarily guides the semi-finished product to an isolation buffer position and processes it first when the robot arm is idle. If neither a robot arm nor an isolation position is available, it is at least prohibited from entering the critical downstream station to avoid contaminating subsequent cycles. For example, during the night shift of headlight assembly, at 23:41, a headlight housing that has been coated with glue stayed in the buffer section for more than a preset time threshold due to congestion of the upstream AGV; the system calculated that its tensor norm reached 0.79, which is higher than the critical solidification threshold of 0.75; at this time, although the standby curing oven has just released a position, the system does not continue to plan a path for it, but directly controls the sorting robot arm to grab it at the end of the buffer section and put it into the scrap isolation rack; the system releases the workstation resources originally allocated to this semi-finished product to the next semi-finished product with a tensor norm lower than the critical solidification threshold; By moving the scrap determination process forward to before the scheduling stage, we can isolate semi-finished products that have failed in physical and chemical processes in advance, reduce unnecessary processing loads, and avoid the equipment at subsequent workstations being idle. Also includes: Real-time monitoring of the fault status of each workstation on the automated production line; If a fault is detected at any workstation in the automated production line, the workstation with the fault is designated as an abnormal workstation. Based on the gradient difference from the abnormal workstation to its upstream node in the continuous gradient field, the reverse transmission pressure value is obtained by multiplying the gradient difference with the preset distance attenuation coefficient. The reverse transmission pressure value is compared with the preset pressure threshold; If the reverse transmission pressure value is greater than or equal to the pressure threshold, a speed reduction command is generated, and the processing equipment corresponding to the station upstream of the abnormal station is controlled to reduce the cycle time parameter. If the reverse transmission pressure value is less than the pressure threshold, the current cycle parameters of the processing equipment are maintained.

[0025] This embodiment provides a reverse pressure transmission control mechanism for fault conditions. Specifically, in the aforementioned process, if the current work-in-process is bypassed only after a fault occurs without adjusting the cycle time of the upstream equipment, high-risk semi-finished products will continue to be generated and concentrate in the obstructed area, eventually causing the buffer capacity limit to be exceeded and the material flow of the entire line to be obstructed. Therefore, after detecting the abnormal workstation, this embodiment pushes the pressure upstream based on the gradient difference in the continuous gradient field, and actively reduces the upstream processing cycle time when the pressure exceeds the threshold. The details are as follows: The system continuously reads fault status signals from each workstation, such as emergency stop, alarm, door access control not closed, or driver fault position from the PLC; if a fault is detected in a UV curing oven workstation F, it is marked as an abnormal workstation, and the gradient difference from that workstation to each upstream node is extracted; for example, if the local gradient value of abnormal workstation F is 0.90, the upstream node U1 is 0.55, and the upstream node U2 is 0.40, then the corresponding gradient differences are 0.35 and 0.50, respectively. A preset distance attenuation coefficient is introduced to reflect that the impact of a fault weakens as the topological distance increases. Assuming the distance attenuation coefficient from the abnormal workstation F to U1 is 0.9 and to U2 is 0.7, the reverse transmission pressure values ​​are 0.35×0.9=0.315 and 0.50×0.7=0.35, respectively. If the system pressure threshold is set to 0.30, both reach the threshold, indicating that the impact of the fault is sufficient to suppress the upstream output. Based on this, the system issues speed reduction commands to the corresponding equipment of U1 and U2, for example, reducing the cycle time of the glue coating robot from 12 seconds per piece to 15 seconds per piece, and reducing the cycle time of the injection molding line from 20 seconds per piece to 24 seconds per piece. If the gradient difference of a more distant upstream node U3 is 0.28 and the distance decay coefficient is 0.6, then the pressure value is 0.168, which is lower than the threshold. At this time, U3 can maintain the current cycle time and does not need to participate in the deceleration. Through this hierarchical control, the entire production line can avoid adopting a unified global deceleration strategy. Compared to simply bypassing faults, this mechanism can form a self-balancing closed loop; the fault point is no longer just a trigger point for local path changes, but also becomes a feedback source for adjusting the overall production cycle; this can prevent upstream from continuing to deliver semi-finished products at full production speed, causing high-risk parts to accumulate disorderly in the buffer section. Regarding the anomaly handling mechanism, if the fault status fluctuates, for example, if workstation F repeatedly recovers from the fault within a period shorter than the preset time interval, the system can set a minimum confirmation time. Only if the fault continues for more than the preset time will the reverse pressure calculation be triggered. If multiple workstations fail simultaneously, their respective reverse transmission pressure values ​​can be calculated separately and then superimposed by nodes to avoid missing compound effects. If the speed reduction command fails to be issued, the system records the failed workstation and raises the downstream buffer warning level of that workstation. If the upstream equipment is already at the minimum safe cycle time, the speed reduction will not continue, but the priority will be to suspend material feeding or take over manually. For example, in the mixed production line of headlights, at 00:18, the main UV curing oven stopped abnormally due to an interlocking problem. After detecting the fault, the system found that although some semi-finished products downstream had been diverted, the upstream gluing robot was still working at its normal cycle time. Based on the gradient difference between the abnormal station and the gluing outlet buffer section, as well as the distance attenuation coefficient between the two, the system calculated that the reverse transmission pressure reached 0.34, exceeding the threshold of 0.30. Therefore, the system automatically reduced the cycle time of the gluing robot and simultaneously adjusted the cycle time of the upstream injection molding line. As a result, the production line maintained controlled operation even when the fault was not completely eliminated, instead of continuous material accumulation. The impact of the faulty workstation is expanded from a local stagnation point to a quantifiable reverse control signal, thereby realizing the dynamic contraction of the upstream cycle time and avoiding the amplification of the abnormality throughout the entire production line; Obtain the current position node, physical environment parameters, and dwell time of the semi-finished vehicle light product at the current workstation corresponding to the current position node in a continuous process, including: Get the current location node and the dwell time; The temperature and humidity sensor array is controlled by the edge computing gateway to collect data according to a preset sampling period; wherein the preset sampling period is less than or equal to fifty milliseconds. Based on the data acquisition results, ambient temperature and humidity data were extracted. Ambient temperature and humidity data are used as physical environmental parameters.

[0026] This embodiment provides an edge-side high-frequency acquisition mechanism. Specifically, in the aforementioned scheduling process, if the sampling frequency of environmental parameters is lower than a preset frequency threshold, such as updating at the second or even minute level, then the state changes in the short critical window from after applying the adhesive to before curing will not be captured in time. Especially under the combined effects of local heat sources, door opening and closing, and workstation congestion, temperature and humidity will change beyond the allowable deviation range within a period lower than the set time threshold. Therefore, this embodiment uses an edge computing gateway to control the temperature and humidity sensor array to collect data with a sampling period of no more than fifty milliseconds. The details are as follows: The system acquires the current location node of the semi-finished product and its dwell time at the current workstation; the current location node can be determined by a radio frequency identifier, a conveyor section encoder, or an AGV positioning tag; the dwell time can be obtained by subtracting the current time from the timestamp of the semi-finished product entering the current node; the edge computing gateway sends sampling control signals to the temperature and humidity sensor array deployed in the adjacent area; the sensor array does not only contain a single-point sampler, but also distributes multiple sampling points along key areas such as the glue coating outlet, buffer section, and curing oven inlet to avoid sampling deviations caused by local airflow; To demonstrate its operation, assume three temperature and humidity sensors are deployed in a certain buffer segment. At 50 milliseconds, the temperatures are 29.8, 30.1, and 29.9 degrees Celsius, and the humidity is 62%, 63%, and 62%, respectively. At the next sampling period, 100 milliseconds, the temperatures are 30.0, 30.5, and 30.2 degrees Celsius, and the humidity is 63%, 65%, and 64%, respectively. The edge computing gateway can use local averaging or weighted averaging to generate the ambient temperature and humidity data for this node. For example, the fused result at 100 milliseconds could be a temperature of 30.23 degrees Celsius and a humidity of 64%. This result is used as physical environment parameters input into the subsequent tensor calculation module. Using an edge computing gateway instead of uploading all the raw data to the central server helps reduce transmission latency. Especially when there are frequent workstation switching and multiple batches running in parallel, the edge side can first complete time alignment, outlier removal and node-level fusion, and then send the simplified results to the scheduling center. In the anomaly handling mechanism, if a sensor does not return data within a sampling period, its value from the previous period can be used, and the reliability of the node's data can be reduced. If more than half of the sensors in the same area fail, the system marks the node as an environmental sampling degradation zone, and subsequent scheduling will prioritize a more conservative path. If the return values ​​of multiple sensors differ too much, for example, if the temperature difference exceeds 3 degrees Celsius at the same time, the edge computing gateway can remove outliers based on spatial proximity. If the network is temporarily interrupted, the edge computing gateway can still cache the sampling data of several periods locally and perform short-term autonomous calculations, and then retransmit it after the connection is restored. For example, in the production of night-shift headlights, there is a maintenance door that is frequently opened near the glue-coating buffer section; at 01:07 AM, during the short period of time the door is open, the sensor on the side near the door detects that the rate of increase in humidity exceeds the preset change threshold. The edge computing gateway captures this change within a 50-millisecond cycle and merges it to show that the humidity of the node has increased from 61% to 67%. Based on this, the scheduling system judges that the decay rate of the semi-finished products in this area will exceed the preset decay standard value, so it raises its processing priority in advance, instead of waiting for the central server to refresh periodically before passively processing it. This provides a timely and reliable data foundation for subsequent physicochemical state mapping, thereby enabling timely response to short-term changes in operating conditions and avoiding scheduling misjudgments caused by sampling lag. Routing and diversion instructions are generated based on the steepest descent path to control the entry of the semi-finished vehicle headlights into the target workstation, including: Determine the target direction based on the steepest descent path; The target direction is encapsulated into routing and routing commands; The routing command is sent to the automated guided vehicle (AGV) scheduling interface configured on the automated production line to control the AGVs configured on the automated production line to move to the target direction; The diversion command is sent to the diversion baffle actuator configured in the automated production line to control the conveyor belt diversion baffle configured in the automated production line to switch to the target direction.

[0027] This embodiment provides an instruction encapsulation and delivery mechanism for the execution layer. Specifically, in the aforementioned scheduling process, even if the steepest descent path has been calculated, if the path cannot be converted into a control command that the underlying device can recognize, the continuous gradient field remains at the analysis level and cannot form a closed loop. Therefore, this embodiment further maps the steepest descent path to the target direction and encapsulates it into AGV routing instructions and conveyor belt diversion instructions, which are then delivered to the corresponding execution interface. The details are as follows: The system determines the target direction based on the steepest descent path; the target direction here can be a discrete control direction such as entering the backup curing oven C from the left branch, going straight into the main curing oven B, turning right into the isolation buffer position; assuming the current position node is the first backup curing oven, and the path obtained through path finding is the first backup curing oven → J4 → C, then this path can be simplified and extracted as the flow is diverted at the first backup curing oven to the right branch, and then sent by the AGV to the entrance C; The system encapsulates the target direction into executable instructions. For a simplified example, a diversion instruction can include a node number, target branch number, action time limit, and receipt identifier. For instance, a node for the first backup curing oven can switch to branch R with a 300-millisecond completion time limit. A routing instruction can include the AGV number, starting point, target point, and priority. For example, AGV07 can retrieve parts from point P3, target entry point C, and has a high priority. These encapsulated instructions can be directly adapted to the underlying interface without requiring manual interpretation. The system sends routing instructions to the automated guided vehicle (AGV) dispatching interface. The AGV dispatching interface allocates specific vehicles based on the current availability of vehicles, path occupancy, and battery status. If the nearest available vehicle is AGV07, it is controlled to drive to the pickup point. If AGV07 has insufficient battery power, the dispatching interface can reassign AGV09. At the same time, the system sends diversion instructions to the diversion baffle actuator, so that the conveyor belt completes the baffle switching in advance before the semi-finished product reaches the branch point. After the baffle is executed, it can send the arrival signal back to the dispatching system as a prerequisite for whether the AGV is allowed to enter the site to pick up the product. In some scenarios, the target direction only involves the diversion baffle, and no AGV is needed; in other scenarios, only AGV transportation is required, and there is no conveyor belt diversion. The system can automatically cut the instruction type according to the path structure. If the path involves both, the baffle and the conveyor path are made ready first before the AGV is released, so as to avoid the blockage caused by the channel not being switched when it arrives. In the exception handling mechanism, if the automated guided vehicle (AGV) dispatch interface returns that no vehicles are available, the system can decide whether to switch to the nearest conveyor branch or enter a high-priority buffer position based on the current tensor value of the semi-finished product. If the diversion baffle actuator does not return a signal within the specified time limit, the system will prohibit the semi-finished product from entering the bifurcation section and recalculate the alternative path. If the AGV encounters a temporary traffic conflict during its journey, the dispatch interface can report the estimated delay time upwards, and the system will re-evaluate whether the path still meets the physical and chemical window requirements. If not, the current task will be canceled and reassigned to another target workstation. For example, during the night shift production of headlights, at 02:26 AM, the system calculates the path buffer segment for a high-risk lamp housing: first backup curing oven → right branch → backup curing oven C. The system issues a right turn command to the diversion baffle at the first backup curing oven and sends a pickup task to the automated guided vehicle (AGV) scheduling interface. After the baffle returns a positioning signal within 180 milliseconds, AGV07 enters the first backup curing oven to pick up the part and drives along the predetermined path to entrance C. If AGV07 exceeds the expected delay limit due to meeting oncoming traffic at an intersection, the scheduling system can immediately cancel the task and have it taken over by the closer AGV11 to ensure that the lamp housing enters the curing process before physical and chemical failure. The upper-level pathfinding results are seamlessly transformed into lower-level executable control actions, thereby achieving closed-loop linkage between scheduling analysis and on-site execution, ensuring that the continuous gradient field truly affects the actual flow of the vehicle lamp semi-finished products.

[0028] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. An automated manufacturing execution system for automotive lighting production lines, characterized in that, include: The edge data acquisition module is used to acquire in real time the current position node, physical environment parameters, and dwell time of the semi-finished automotive lamp in the continuous process at the current workstation corresponding to the current position node; wherein, the continuous process includes manufacturing links with irreversible physical and chemical property decay characteristics. The spatiotemporal rheological tensor calculation module is used to input physical environment parameters and residence time into a preset decay dynamics equation and map them into a dynamic viscosity tensor; among which, the dynamic viscosity tensor is used to characterize the physicochemical decay state of the semi-finished vehicle lamp. The gradient field scheduling module is used to map the equipment operating efficiency value of each workstation in the automated production line as a depth parameter to a gravitational potential well that represents the scheduling priority of the workstation, and to construct the topological continuous gradient field corresponding to the automated production line based on the dynamic viscosity tensor and the gravitational potential well. The underlying execution module is used to extract the steepest descent path to the preset target workstation based on the continuous gradient field, starting from the current position node, and generate routing instructions and diversion instructions based on the steepest descent path and send them to the corresponding execution devices to control the semi-finished vehicle lights to enter the target workstation.

2. A production scheduling method for automated production lines of automotive lighting, characterized in that, The manufacturing execution system for an automated production line of automotive lighting as described in claim 1 includes: Obtain the current position node, physical environment parameters, and dwell time of the semi-finished vehicle light product at the current workstation corresponding to the current position node in a continuous process. The physical environment parameters and residence time are input into the preset decay dynamics equation and mapped to a dynamic viscosity tensor. The equipment operating efficiency value of each workstation in the automated production line is used as a depth parameter and mapped to a gravitational potential well that represents the scheduling priority of the workstation. Based on the dynamic viscosity tensor and the gravitational potential well, a topological continuous gradient field corresponding to the automated production line is constructed. Based on the continuous gradient field, starting from the current position node, extract the steepest descent path to the preset target workstation; Routing and diversion instructions are generated based on the steepest descent path to control the entry of semi-finished vehicle lights into the target workstation.

3. The production scheduling method for automated production lines of automotive lights according to claim 2, characterized in that, The physical environment parameters and residence time are input into a preset decay kinetic equation, which is mapped to a dynamic viscosity tensor, including: Extract pre-defined order data from the manufacturing execution system; Initialize the rheological tensor baseline of the semi-finished automotive headlights based on order data from the Manufacturing Execution System; Input the physical environment parameters and residence time into the preset decay kinetic equation, calculate the product of the deviation coefficient of the physical environment parameters relative to the reference process parameters and the residence time, and multiply it by the preset decay weight matrix to obtain the decay evolution value of the matrix dimension. The rheological tensor baseline and the decay evolution value are fused by matrix addition to generate a dynamic viscosity tensor.

4. The production scheduling method for automated production lines of automotive lights according to claim 2, characterized in that, The equipment operating efficiency values ​​of each workstation in the automated production line are used as depth parameters and mapped to gravitational potential wells representing workstation scheduling priorities. A topological continuous gradient field corresponding to the automated production line is constructed based on the dynamic viscosity tensor and the gravitational potential wells, including: The overall equipment efficiency feedback value of each workstation in the automated production line is obtained through a programmable logic controller and used as the equipment operating efficiency value. The reciprocal of the overall equipment efficiency feedback value is used to determine the potential depth of the gravitational potential well. The potential well depth is multiplied by the dynamic viscosity tensor to generate a viscosity gradient field. The viscosity gradient field is output as a continuous gradient field.

5. The production scheduling method for automated production lines of automotive lighting according to claim 2, characterized in that, Based on the continuous gradient field, before extracting the steepest descent path to the preset target workstation, starting from the current position node, the process also includes: Calculate the matrix norm of the dynamic viscosity tensor and compare it with the preset critical solidification threshold; If the matrix norm is greater than or equal to the critical solidification threshold, a scrap rejection instruction is generated and sent to the corresponding robotic arm controller to control the sorting robotic arm configured in the automated production line to remove the semi-finished car light product from the automated production line. If the matrix norm is less than the critical solidification threshold, then the step of extracting the steepest descent path to the preset target workstation based on the continuous gradient field and starting from the current position node is executed.

6. The production scheduling method for automated production lines of automotive lights according to claim 2, characterized in that, Also includes: Real-time monitoring of the fault status of each workstation on the automated production line; If a fault is detected at any workstation in the automated production line, the workstation with the fault is designated as an abnormal workstation. Based on the gradient difference from the abnormal workstation to its upstream node in the continuous gradient field, the reverse transmission pressure value is obtained by multiplying the gradient difference with the preset distance attenuation coefficient. The reverse transmission pressure value is compared with the preset pressure threshold; If the reverse transmission pressure value is greater than or equal to the pressure threshold, a speed reduction command is generated, and the processing equipment corresponding to the station upstream of the abnormal station is controlled to reduce the cycle time parameter. If the reverse transmission pressure value is less than the pressure threshold, the current cycle parameters of the processing equipment are maintained.

7. The production scheduling method for automated production lines of automotive lights according to claim 2, characterized in that, Obtain the current position node, physical environment parameters, and dwell time of the semi-finished vehicle light product at the current workstation corresponding to the current position node in a continuous process, including: Get the current location node and the dwell time; The temperature and humidity sensor array is controlled by the edge computing gateway to collect data according to a preset sampling period; wherein the preset sampling period is less than or equal to fifty milliseconds. Based on the data acquisition results, ambient temperature and humidity data were extracted. Ambient temperature and humidity data are used as physical environmental parameters.

8. The production scheduling method for automated production lines of automotive lights according to claim 2, characterized in that, Routing and diversion instructions are generated based on the steepest descent path to control the entry of the semi-finished vehicle headlights into the target workstation, including: Determine the target direction based on the steepest descent path; The target direction is encapsulated into routing and routing commands; The routing command is sent to the automated guided vehicle (AGV) scheduling interface configured on the automated production line to control the AGVs configured on the automated production line to move to the target direction; The diversion command is sent to the diversion baffle actuator configured in the automated production line to control the conveyor belt diversion baffle configured in the automated production line to switch to the target direction.

Images