Digital process control method and application of double-extraction skin care capsule production line

Abstract

The invention provides a digital process control method and application of a double-extraction skin care capsule production line, and relates to the technical field of production control, and the method comprises the steps: obtaining the real-time operation state data of a plurality of devices; extracting process stage identifiers and constraint conditions; constructing a dynamic collaborative topological structure; determining a bidirectional cooperative constraint relationship between the devices; converting the constraint into a condition set of an optimization problem; and solving an optimal control parameter combination by adopting a heuristic search and gradient optimization algorithm. According to the invention, the process waiting time between equipment can be minimized, the material circulation continuity can be maximized, and the production efficiency and the product quality stability can be improved.

Description

Technical Field

[0001] This invention relates to the field of production control technology, and in particular to a digital process control method and application for a double-extract skincare capsule production line. Background Technology

[0002] Double-extract skincare capsules are high-end products in the modern beauty and health industry. Their production process involves multiple steps, including material formulation, emulsification and mixing, and capsule forming. As consumers' demands for skincare product quality increase, the production process of double-extract skincare capsules is becoming increasingly complex, placing higher demands on the precise control of the production line. Traditional double-extract skincare capsule production lines mainly rely on manual operation and simple single-machine automation control, lacking effective information exchange and collaborative control mechanisms between production equipment.

[0003] With the development of industrial internet technology, digital control technology has begun to be introduced into the skincare product manufacturing industry. Currently, some companies are attempting to apply intelligent control systems to dual-extract skincare capsule production lines to improve production efficiency and product quality. These systems typically control individual machines independently based on preset process parameters or use simple sequential logic to achieve basic inter-machine linkage. However, existing technologies still have significant shortcomings in the digital process control of dual-extract skincare capsule production lines.

[0004] Existing control systems lack the ability to dynamically perceive the process status of production equipment in real time, making it difficult to accurately identify the process stages and constraints of each piece of equipment. This leads to problems such as mismatched process parameters and large quality fluctuations during production. In particular, when raw material characteristics change or equipment status fluctuates, it is impossible to adjust relevant process parameters in a timely manner, affecting the stability of product quality.

[0005] Traditional control methods fail to effectively establish collaborative relationship models between equipment, and cannot accurately describe the process connections and constraints during material flow. Equipment often exists in "information silos," lacking effective status information sharing and collaborative decision-making mechanisms, resulting in low overall production line efficiency, long material transfer waiting times, and significant energy waste. Summary of the Invention

[0006] This invention provides a digital process control method and application for a dual-extract skincare capsule production line, which can solve the problems in the prior art.

[0007] A first aspect of this invention provides a digital process control method for a double-extract skincare capsule production line, comprising: The system acquires real-time operating status data of multiple production devices in the dual-extract skincare capsule production line, parses the real-time operating status data, and extracts the current process stage identifier and corresponding process constraints of each production device. Based on the process stage identifiers, a dynamic collaborative topology is constructed that reflects the material transfer relationship and process connection relationship between equipment. The state transition conditions of each equipment node and the material transfer constraints of the edge connections are marked in the dynamic collaborative topology. Based on the topological connectivity and node state distribution of the dynamic collaborative topology, the bidirectional collaborative constraint relationship between each production equipment is determined. The bidirectional collaborative constraint relationship is transformed into a set of constraints for a multi-equipment collaborative optimization problem, and the process constraints corresponding to each production equipment are transformed into the boundary values ​​of decision variables. A heuristic search algorithm is used to explore feasible solution regions in the decision space that satisfy the set of constraints and generate an initial solution set. A gradient optimization algorithm is then used to perform local optimization within the feasible solution region, starting from the initial solution set, to solve the multi-equipment collaborative optimization objective function with the optimization objectives of minimizing the process waiting time between equipment and maximizing the continuity of material flow, thereby obtaining the optimal combination of control parameters for each production equipment.

[0008] Based on the process stage identifiers, a dynamic collaborative topology is constructed that reflects the material transfer relationships and process connection relationships between equipment. The state transition conditions of each equipment node and the material transfer constraints of the edge connections are marked in the dynamic collaborative topology, including: Each production device is mapped to a device node in the dynamic collaborative topology, and the material transfer path between devices is mapped to a directed edge connecting the device nodes; According to the process stage identifier of each production equipment, the current process state is marked on the corresponding equipment node. According to the process flow rules, the state transition conditions are marked on each equipment node. The state transition conditions define the prerequisites that the equipment node must meet to transition from the current process state to the next process state. Material transfer constraints are marked on each directed edge. The material transfer constraints define the material state requirements and transfer timing requirements that must be met when the source device node transfers materials to the target device node.

[0009] Based on the topological connectivity and node state distribution of the dynamic collaborative topology, the bidirectional collaborative constraint relationships between various production devices are determined, including: Based on the topological connectivity of the dynamic collaborative topology, each directed edge in the dynamic collaborative topology is traversed, and the source device node and target device node corresponding to each directed edge are identified; the state transition constraints related to material output are extracted from the state transition conditions marked by the source device node, and the material transfer constraints are extracted from the directed edges connecting the source device node and the target device node; The state transition constraints and the material transfer constraints are logically conjuncted to obtain the forward coordination constraints between the source device node and the target device node; the state transition constraints related to material receiving are extracted from the state transition conditions marked by the target device node, and these state transition constraints are used as the reverse coordination constraints of the target device node on the source device node; The forward collaborative constraint and the reverse collaborative constraint are combined to form a bidirectional collaborative constraint relationship.

[0010] The heuristic search algorithm is used to explore feasible solution regions in the decision space that satisfy the set of constraints and generate an initial solution set, including: The dimensions and boundary range of the decision space are defined based on the set of constraints. Multiple candidate solutions are randomly generated in the decision space, and each candidate solution corresponds to a set of control parameter values ​​for each production equipment. Each candidate solution is subjected to constraint verification to determine whether it satisfies all constraints in the constraint set. Candidate solutions that satisfy all constraints are marked as feasible solutions. A heuristic search algorithm is used to perform a neighborhood search in the decision space based on the feasible solution. New candidate solutions are generated by perturbing the control parameters of the feasible solution, and the constraints of the new candidate solutions are verified. The new candidate solutions that pass the verification are added to the feasible solution set. The neighborhood search process is repeated until the size of the feasible solution set reaches the preset size requirement. The feasible solution set is then used as the initial solution set.

[0011] Using the gradient optimization algorithm, starting from the initial solution set, local optimization is performed within the feasible solution region to solve the multi-equipment collaborative optimization objective function, which aims to minimize the process waiting time between equipment and maximize the continuity of material flow. The optimal control parameter combinations for each production equipment are obtained, including: The initial solution with the optimal objective function value for multi-device collaborative optimization is selected from the initial solution set as the starting point for gradient optimization; The multi-equipment collaborative optimization objective function is decomposed into a sub-objective function that minimizes the inter-equipment process waiting time and a sub-objective function that maximizes the continuity of material flow. The gradient vectors of the sub-objective functions that minimize the inter-equipment process waiting time and the sub-objective functions that maximize the continuity of material flow are calculated at the starting point to obtain the first sub-objective gradient vector and the second sub-objective gradient vector. Calculate the angle between the gradient vectors of the first sub-target and the second sub-target, and determine the consistency of the optimization directions of the two sub-targets. When the angle is less than a preset angle threshold, the weighted sum of the gradient vectors of the first and second sub-targets is used as the coordinating gradient direction. When the angle is greater than the preset angle threshold, the Pareto front projection method is used to project the gradient vectors of the two targets onto the Pareto improvement direction to obtain the coordinating gradient direction. The control parameters of the starting point are adjusted along the direction of the coordinated gradient. The parameter adjustment step size is determined by a line search method, and the adjusted control parameters are combined as a new starting point. Repeated iterations are performed. When the magnitudes of the gradient vectors of the first and second sub-targets are both less than the preset convergence threshold after multiple consecutive iterations, the combination of control parameters corresponding to the current starting point is taken as the optimal combination of control parameters for each production device.

[0012] The Pareto front projection method is used to project the two target gradient vectors onto the Pareto improvement direction to obtain the coordinated gradient direction, which includes: A Pareto improvement cone is constructed, which is defined by the set of all search directions that simultaneously decrease the sub-objective function values ​​corresponding to the gradient vectors of the first and second sub-objectives. The coordinated gradient direction is determined by solving the optimal projection direction of the weighted combination vector of the first sub-objective gradient vector and the second sub-objective gradient vector within the Pareto improved cone. The optimal projection direction is obtained by maximizing the magnitude of the projection vector within the Pareto improved cone while simultaneously satisfying the constraint condition that both sub-objective function values ​​decrease. Calculate the dot product of the coordinated gradient direction and the gradient vector of the first sub-target, and calculate the dot product of the coordinated gradient direction and the gradient vector of the second sub-target to verify that the coordinated gradient direction simultaneously satisfies the descent requirements of both sub-targets; The verified coordinated gradient direction will be used as the search direction for subsequent parameter adjustments.

[0013] The method also includes constructing a sub-objective function that minimizes process waiting time between equipment and a sub-objective function that maximizes the continuity of material flow: Traverse each directed edge in the dynamic cooperative topology and identify the source device node and target device node corresponding to each directed edge; The material output completion time is determined based on the source device node, and the material receiving readiness time is determined based on the target device node. The time difference between the material receiving readiness time and the material output completion time is used as the inter-device waiting time corresponding to the directed edge. The waiting times between devices corresponding to all directed edges in the dynamic cooperative topology are summed to obtain the total waiting time of the overall inter-device process waiting loss. The total waiting time is used as the optimization objective of the sub-objective function for minimizing the inter-device process waiting time. Traverse each directed edge in the dynamic cooperative topology, determine the material transfer timing conditions from the material transfer constraints marked on each directed edge, and determine whether the material output time of the source device node and the material receiving time of the target device node satisfy the material transfer timing conditions; The number of directed edges that satisfy the material transfer timing conditions is counted, and the ratio of the number of directed edges that satisfy the material transfer timing conditions to the total number of directed edges is calculated as the material transfer timing compliance rate. Traverse each device node in the dynamic collaborative topology, extract the material batch sequence processed by each device node, and count the number of process state switching between adjacent material batches as the process switching count of the device node; The total number of process switching times is obtained by summing the number of process switching times of all device nodes in the dynamic collaborative topology. The material transfer timing compliance rate is weighted and summed with the reciprocal of the total number of process switching times to obtain a flow continuity index that characterizes the continuity of material flow. The flow continuity index is used as the optimization objective of the sub-objective function for maximizing the continuity of material flow.

[0014] A second aspect of the present invention provides a dual-extract oil essence capsule, comprising: The whitening capsule formula and the anti-aging capsule formula are prepared in parts by weight, and the whitening capsule formula and the anti-aging capsule formula are filled into the same capsule in the same volume ratio; The whitening formula comprises the following components and parts by weight: 30-40 parts of coconut oil alcohol-caprylate / capric acid ester, 32-36 parts of isododecane, 11-18 parts of isohexadecane, 2-5 parts of hydrogenated styrene / isoprene copolymer, 3.5-5.5 parts of pentaerythritol tetrabis-tert-butylhydroxyhydrogenated cinnamic acid ester, 2.5-4.5 parts of oil-soluble niacinamide-containing whitening active ingredient combination, and 0 parts of phenylethyl resorcinol-containing whitening active ingredient combination. 5-1.5 parts, a whitening active ingredient combination containing tranexamic acid and nonapeptide-1 (0.2-0.6 parts), bisabolol (0.2-0.6 parts), citrus essential oil blend (0.15-0.5 parts), myristyl nicotinate (0.05-0.25 parts), palm bark extract (0.08-0.12 parts), glycyrrhizin (0.08-0.12 parts), stearyl glycyrrhetinic acid ester (0.04-0.12 parts), and tocopherol (0.04-0.12 parts); The anti-aging capsule formulation comprises the following components and parts by weight: 30-45 parts of cyclopentamethoxysiloxane, 24-40 parts of polydimethylsiloxane, 3-6.5 parts of hexyldecyl alcohol, 3-6 parts of octyl polymethylsiloxane, 2.5-4.5 parts of divinyl polydimethylsiloxane / polydimethylsiloxane crosspolymer, 2.5-4.5 parts of polydimethylsiloxane / vinyl polydimethylsiloxane crosspolymer, 2-3 parts of coconut oil alcohol-caprylate / caprylate, and 0.9 parts of phytosterol / octyldodecyl lauroyl glutamate. -2.2 parts, 0.4-1.2 parts of anti-aging active ingredient combination containing hybrid rose flower extract, 0.2-0.6 parts of anti-aging active ingredient combination containing sodium DNA and ergothioneine, 0.3-0.5 parts of citrus essential oil compound, 0.15-0.5 parts of bisabolol combination containing ginger root extract, 0.04-0.12 parts of rosewood bark extract, 0.04-0.12 parts of stearyl glycyrrhetinic acid ester, 0.04-0.12 parts of tocopherol, and 0.0001-0.001 parts of sugar isomer combination.

[0015] A third aspect of the present invention provides a digital process control system for a dual-extract skincare capsule production line, comprising: The first unit is used to acquire real-time operating status data of multiple production equipment in the dual-extract skincare capsule production line, parse the real-time operating status data, and extract the current process stage identifier and corresponding process constraints of each production equipment; The second unit is used to construct a dynamic collaborative topology structure that reflects the material transfer relationship and process connection relationship between equipment based on the process stage identifier. In the dynamic collaborative topology structure, the state transition conditions of each equipment node and the material transfer constraints of the edge connection are marked. Based on the topological connectivity and node state distribution of the dynamic collaborative topology structure, the bidirectional collaborative constraint relationship between each production equipment is determined. The third unit is used to transform the bidirectional collaborative constraint relationship into a set of constraints for a multi-device collaborative optimization problem, and to transform the process constraints corresponding to each production device into the boundary values ​​of decision variables; The fourth unit is used to explore feasible solution regions in the decision space that satisfy the set of constraints using a heuristic search algorithm and generate an initial solution set. Then, using a gradient optimization algorithm, it performs local optimization within the feasible solution region starting from the initial solution set to solve the multi-device collaborative optimization objective function with the optimization objectives of minimizing the process waiting time between equipment and maximizing the continuity of material flow, and obtains the optimal combination of control parameters for each production equipment.

[0016] Fourth aspect of the embodiments of the present invention, A computer-readable storage medium is provided, having stored thereon computer program instructions that, when executed by a processor, implement the aforementioned method.

[0017] The beneficial effects of this application are as follows: This application constructs a dynamic collaborative topology that reflects the material transfer relationship and process connection relationship between equipment. This structure can adapt to changes in production status, accurately characterize the complex interaction relationship between equipment, and provide a mathematical basis for multi-equipment collaborative optimization.

[0018] Transforming complex bidirectional collaborative constraint relationships into a standardized set of constraints enables high-dimensional, strongly coupled production control problems to be effectively solved, significantly improving the problem's tractability.

[0019] A hybrid optimization strategy combining heuristic search and gradient optimization is adopted, which not only ensures global search capability and avoids getting trapped in local optima, but also uses gradient information to accelerate the convergence process, significantly improving the solution efficiency and solution quality.

[0020] By optimizing multiple objectives—minimizing process waiting time between equipment and maximizing material flow continuity—material backlog and equipment idleness in traditional production lines are effectively eliminated, significantly improving the collaborative operation efficiency and capacity utilization of the production line. Attached Figure Description

[0021] Figure 1 This is a flowchart illustrating the digital process control method for the dual-extract skincare capsule production line according to an embodiment of the present invention. Figure 2 A flowchart illustrating the process of obtaining the optimal combination of control parameters for each production device. Detailed Implementation

[0022] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0023] The technical solution of the present invention will be described in detail below with reference to specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments.

[0024] refer to Figure 1 and Figure 2 The digital process control method for a dual-extract skincare capsule production line according to an embodiment of the present invention includes: The system acquires real-time operating status data of multiple production devices in the dual-extract skincare capsule production line, parses the real-time operating status data, and extracts the current process stage identifier and corresponding process constraints of each production device. Based on the process stage identifiers, a dynamic collaborative topology is constructed that reflects the material transfer relationship and process connection relationship between equipment. The state transition conditions of each equipment node and the material transfer constraints of the edge connections are marked in the dynamic collaborative topology. Based on the topological connectivity and node state distribution of the dynamic collaborative topology, the bidirectional collaborative constraint relationship between each production equipment is determined. The bidirectional collaborative constraint relationship is transformed into a set of constraints for a multi-equipment collaborative optimization problem, and the process constraints corresponding to each production equipment are transformed into the boundary values ​​of decision variables. A heuristic search algorithm is used to explore feasible solution regions in the decision space that satisfy the set of constraints and generate an initial solution set. A gradient optimization algorithm is then used to perform local optimization within the feasible solution region, starting from the initial solution set, to solve the multi-equipment collaborative optimization objective function with the optimization objectives of minimizing the process waiting time between equipment and maximizing the continuity of material flow, thereby obtaining the optimal combination of control parameters for each production equipment.

[0025] For example, the dual-extract skincare capsule production line includes raw material pretreatment equipment, mixing and stirring equipment, filling and packaging equipment, quality inspection equipment, and packaging conveying equipment. Each piece of production equipment is equipped with data acquisition sensors, which communicate with the central control system via a fieldbus protocol. The data acquisition sensors include temperature sensors, pressure sensors, flow sensors, speed sensors, and position sensors. The temperature sensor monitors the temperature changes inside the equipment in real time, with a sampling frequency set to 10 times per second, and the accuracy of the temperature value obtained for each sample reaches 0.1 degrees Celsius. The pressure sensor monitors the pressure status inside the mixing and stirring equipment, triggering an early warning signal when the pressure value exceeds the set range. The flow sensor is installed in the material conveying pipeline and records the volume of material passing through the pipeline per unit time. The speed sensor is connected to the stirring motor shaft, detecting the actual speed of the motor and comparing it with the set speed. The position sensor is used to confirm the precise position of the filling head, ensuring the alignment accuracy between the capsule shell and the filling head.

[0026] The central control system accesses the sensors of each device sequentially via a polling mechanism, with a polling period set to 500 milliseconds. Within each polling period, the system sends a data request command to the target device. Upon response, the device packages the currently collected sensor values ​​into a data frame and returns it to the central control system. The data frame uses a fixed-length format. The frame header contains the device identifier, timestamp, and data type identifier; the frame body contains the specific sensor values; and the frame tail contains a checksum to verify the integrity of the data transmission. Assuming the identifier of the mixing and stirring device is 001, the data frame returned at a certain moment shows a temperature sensor reading of 65.3 degrees Celsius, a pressure sensor reading of 0.8 MPa, and a speed sensor reading of 300 revolutions per minute.

[0027] The extraction of process stage identifiers relies on a predefined process state mapping table. This table maps combinations of sensor values ​​from the equipment to specific process stages. The process stages of a mixing and stirring device include a preheating stage, a feeding stage, a mixing stage, a homogenizing stage, and a holding stage. The preheating stage is characterized by temperature sensor readings between 20 and 50 degrees Celsius, pressure sensor readings close to atmospheric pressure, and a speed sensor reading of 0. The feeding stage is characterized by a flow sensor detecting material inflow with a flow rate greater than 0, a temperature that begins to rise but has not yet reached the required process temperature, and a speed sensor that remains at 0. The mixing stage is characterized by a temperature between 60 and 70 degrees Celsius, a speed sensor reading between 200 and 400 revolutions per minute, and a pressure sensor reading between 0.5 and 1.0 MPa. The homogenizing stage is characterized by a stable temperature near the set value with fluctuations not exceeding ±2 degrees Celsius, a speed increase to 500 to 800 revolutions per minute, and a pressure increase to 1.2 to 1.5 MPa. The characteristics of the heat preservation stage are that the temperature is maintained at the process requirement temperature, the rotation speed is reduced to below 100 revolutions per minute, and the pressure drops back to near atmospheric pressure.

[0028] The system reads the sensor values ​​of the mixing equipment at the current moment: temperature 65.3 degrees Celsius, pressure 0.8 MPa, rotation speed 300 rpm, and flow rate 0. These values ​​are matched with the characteristics of each stage in the process status mapping table. The temperature falls within the range of 60 to 70 degrees Celsius, the pressure falls within the range of 0.5 to 1.0 MPa, the rotation speed falls within the range of 200 to 400 rpm, and the flow rate of 0 indicates that feeding is complete. Based on this comprehensive judgment, the system determines that the equipment is currently in the mixing stage and writes the mixing stage identifier code 003 into the process stage field of the equipment.

[0029] In one optional implementation, based on the process stage identifier, a dynamic collaborative topology structure reflecting the material transfer relationship and process connection relationship between equipment is constructed. The state transition conditions of each equipment node and the material transfer constraints of edge connections are marked in the dynamic collaborative topology structure, including: Each production device is mapped to a device node in the dynamic collaborative topology, and the material transfer path between devices is mapped to a directed edge connecting the device nodes; According to the process stage identifier of each production equipment, the current process state is marked on the corresponding equipment node. According to the process flow rules, the state transition conditions are marked on each equipment node. The state transition conditions define the prerequisites that the equipment node must meet to transition from the current process state to the next process state. Material transfer constraints are marked on each directed edge. The material transfer constraints define the material state requirements and transfer timing requirements that must be met when the source device node transfers materials to the target device node.

[0030] In chemical production processes, constructing a dynamic collaborative topology that reflects the material transfer relationships and process integration relationships between equipment is crucial for optimizing production scheduling. This structure maps production equipment as nodes, material transfer paths as directed edges, and labels state transition conditions and material transfer constraints, enabling visualized management and optimized control of the production process.

[0031] Based on the chemical production process flow diagram and equipment layout diagram, identify all key equipment on the production line, such as reactors, distillation columns, heat exchangers, and storage tanks, and assign a unique equipment number and process stage identifier to each piece of equipment. The process stage identifier typically includes an equipment type code, a process function code, and a serial number, such as "R-101" representing reactor number 1 in the first reaction zone. Equipment type codes can be represented by letters, such as R for reaction equipment, D for separation equipment, and S for storage equipment.

[0032] When mapping each production device to device nodes in a dynamic collaborative topology, a graphical representation method can be used, representing different types of devices with nodes of different shapes or colors. For example, a reactor can be represented as a circular node, a distillation column as a rectangular node, and a storage tank as a hexagonal node. Each node contains information such as the device number, the current process status, and the state transition conditions.

[0033] Material transfer paths between devices are mapped as directed edges connecting device nodes. The direction of the directed edge indicates the direction of material flow, and the thickness of the edge can represent material transfer capacity or transfer priority. For example, in a polymer production line, the material transfer path from the monomer storage tank to the prepolymerization reactor can be represented as a directed edge pointing from the storage tank node to the reactor node.

[0034] The current process status is marked on the corresponding equipment node according to the process stage identifier of each production equipment. Process status can be divided into several types, such as "Idle," "Preparation," "Running," "Heating," and "Cleaning." For critical equipment such as reactors, the process status can be further subdivided into more specific states such as "Feeding," "Heating," "Reaction," "Cooling," and "Discharging." The current process status is displayed through status indicators inside the node, which can be distinguished using different colors or markers.

[0035] When marking state transition conditions on equipment nodes, the prerequisites for transitioning from the current process state to the next process state must be defined according to the process flow rules. State transition conditions typically include three categories: time conditions, material conditions, and equipment conditions. Time conditions specify the duration of the process operation or the transition time; material conditions specify the requirements for parameters such as material composition, temperature, and pressure; equipment conditions specify the requirements for equipment operating parameters, cleanliness, etc.

[0036] For example, the prerequisites for a reactor to transition from the "feeding" state to the "heating" state include: raw materials A, B, and C have been added in the correct proportions; the reactor level has reached the set value; the agitator is operating normally; and the cooling system is in standby mode. These conditions are stored in a structured form in the state transition condition field of the node.

[0037] Material transfer constraints are marked on directed edges, defining the material state requirements and transfer timing requirements that must be met when transferring materials from the source device node to the target device node. Material state requirements include parameters such as the material's temperature range, pressure range, and composition specifications; transfer timing requirements include timing control parameters such as transfer start conditions, transfer end conditions, and transfer rate.

[0038] Material transfer constraints can be represented as edge attributes and displayed through additional information labels. For example, material transfer constraints from the reactor to the crystallizer might include: the material temperature must be controlled at 80±2 degrees Celsius; the transfer flow rate is 2-3 cubic meters per hour; the material pH value should be within the range of 6.5-7.5; and the target equipment must complete cleaning and preheating procedures before the transfer begins.

[0039] When equipment status changes or material transfer needs arise during actual production, the dynamic collaborative topology can be updated in real time. When a piece of equipment completes its current process operation and meets the state transition conditions, the node status is automatically updated to the next process state. Simultaneously, the material transfer path associated with that equipment is dynamically adjusted based on the new process state and material requirements.

[0040] By constructing this dynamic collaborative topology, the collaborative relationships between equipment and material transfer constraints in the chemical production process can be intuitively reflected, providing visualized decision support for production scheduling optimization. For example, when a reactor needs to transfer materials, the system can quickly identify the target equipment that meets the conditions by analyzing the equipment status and material transfer constraints in the topology, and generate the optimal material transfer plan.

[0041] This dynamic collaborative topology also supports emergency handling under abnormal operating conditions. When a piece of equipment malfunctions or its parameters deviate, the system can analyze the scope of impact based on the topology and adjust the process status and material transfer path of the relevant equipment in a timely manner to minimize production interruptions and material losses.

[0042] In one optional implementation, determining the bidirectional collaborative constraint relationship between each production device based on the topological connectivity and node state distribution of the dynamic collaborative topology includes: Based on the topological connectivity of the dynamic collaborative topology, each directed edge in the dynamic collaborative topology is traversed, and the source device node and target device node corresponding to each directed edge are identified; the state transition constraints related to material output are extracted from the state transition conditions marked by the source device node, and the material transfer constraints are extracted from the directed edges connecting the source device node and the target device node; The state transition constraints and the material transfer constraints are logically conjuncted to obtain the forward coordination constraints between the source device node and the target device node; the state transition constraints related to material receiving are extracted from the state transition conditions marked by the target device node, and these state transition constraints are used as the reverse coordination constraints of the target device node on the source device node; The forward collaborative constraint and the reverse collaborative constraint are combined to form a bidirectional collaborative constraint relationship.

[0043] In a smart manufacturing environment, determining the bidirectional collaborative constraint relationships between production equipment based on a dynamic collaborative topology is a key step in achieving efficient collaborative control. By analyzing topological connectivity and node state distribution, an accurate equipment collaborative constraint model can be established, thereby optimizing the production process.

[0044] It is necessary to traverse all directed edges in the dynamic collaborative topology and identify the source and target device nodes corresponding to each directed edge. For example, on a production line, if device A is defined as the source device node and device B as the target device node, and there is a material transfer relationship between them, this is represented in the topology as a directed edge from device A to device B. Through graph traversal algorithms, the systematic identification of all directed edges can be completed, recording the starting and ending device information of each directed edge, laying the foundation for subsequent constraint extraction.

[0045] After identifying the source and target device nodes, state transition constraints related to material output are extracted from the state transition conditions marked on the source device nodes. State transition constraints typically include elements such as device operating status, material status, and operating parameters. For example, for a mixing and stirring device as a source device node, its state transition constraints include the condition "the material temperature reaches the specified value and the mold closure is complete." These constraints directly affect whether the source device can output material to the target device.

[0046] Simultaneously, material transfer constraints are extracted from the directed edges connecting the source and target device nodes. These constraints involve factors such as transmission rate, path availability, and medium status. For example, a conveyor belt between two devices might have material transfer constraints including conditions such as "the conveyor belt speed is ten meters per minute and the load does not exceed fifty kilograms." These constraints determine whether materials can be successfully transferred from the source device to the target device.

[0047] After extracting the state transition constraints and material transfer constraints, a logical conjunction operation is performed on these two types of constraints to obtain the forward coordination constraints between the source device node and the target device node. The logical conjunction operation ensures that the forward coordination constraint is valid only when both the state transition condition and the material transfer condition of the source device are satisfied. For example, a mixing and stirring device can only output the product to the next process when "the material temperature reaches the specified value and the mold is closed" and "the conveyor belt is operating normally and has sufficient load-bearing capacity".

[0048] Next, state transition constraints related to material receiving are extracted from the state transition conditions annotated on the target device node. For example, an assembly machine, as the target device, might have a state transition condition of "receiving area is idle and fixture is in place." These constraints indicate whether the target device can receive material from the source device. This state transition constraint is then used as a reverse collaborative constraint between the target device node and the source device node to control the output behavior of the source device.

[0049] Finally, forward and reverse cooperative constraints are combined to form a bidirectional cooperative constraint relationship. This bidirectional cooperative constraint relationship fully describes all the conditions that must be met during the material transfer process from the source device to the target device, as well as the feedback effect of the target device's state on the source device's behavior. This constraint relationship is dynamic and adjusts with changes in the production environment and equipment status.

[0050] In practical applications, this bidirectional collaborative constraint management can be achieved through a manufacturing execution system (MES). When a constraint is not met, the system automatically adjusts the operating status or parameters of the relevant equipment. For example, when the receiving area of ​​the target equipment is full, the source equipment is notified to suspend output through a reverse collaborative constraint, thereby avoiding material accumulation and production interruption.

[0051] The bidirectional collaborative constraint relationships determined by the above methods can effectively support the collaborative operation of production equipment, reduce production bottlenecks, and improve the overall system efficiency. In complex flexible production lines, this constraint relationship based on a dynamic collaborative topology is particularly important, as it can adapt to frequently changing production demands and process paths.

[0052] Experiments show that the bidirectional collaborative constraint model constructed using this method can improve the material flow efficiency of the production line by 15%-20% and significantly reduce production interruptions caused by improper coordination between equipment.

[0053] In one alternative implementation, exploring feasible solution regions in the decision space that satisfy the set of constraints using a heuristic search algorithm and generating an initial solution set includes: The dimensions and boundary range of the decision space are defined based on the set of constraints. Multiple candidate solutions are randomly generated in the decision space, and each candidate solution corresponds to a set of control parameter values ​​for each production equipment. Each candidate solution is subjected to constraint verification to determine whether it satisfies all constraints in the constraint set. Candidate solutions that satisfy all constraints are marked as feasible solutions. A heuristic search algorithm is used to perform a neighborhood search in the decision space based on the feasible solution. New candidate solutions are generated by perturbing the control parameters of the feasible solution, and the constraints of the new candidate solutions are verified. The new candidate solutions that pass the verification are added to the feasible solution set. The neighborhood search process is repeated until the size of the feasible solution set reaches the preset size requirement. The feasible solution set is then used as the initial solution set.

[0054] The decision space is defined by the given set of constraints, defining its dimensions and boundaries. The dimensions of the decision space are equivalent to the number of control parameters for the production equipment that need to be optimized. These control parameters include multiple parameters such as temperature, pressure, and additive ratios. For each parameter, reasonable upper and lower limits are determined, forming the boundaries of the decision space. These limits are typically determined based on the physical limitations of the equipment and the requirements of the production process.

[0055] Multiple candidate solutions are randomly generated within a defined decision space. Each candidate solution is represented as a vector, with each element corresponding to a control parameter value for a production device. The random generation process employs a uniformly distributed random sampling method to ensure that the initial candidate solutions cover different regions of the decision space. For example, in a three-dimensional decision space, hundreds of random points can be generated, each representing a specific combination of temperature, pressure, and additive ratio.

[0056] After generating candidate solutions, each candidate solution undergoes constraint verification. Constraints are categorized into equality constraints and inequality constraints, and each candidate solution is checked to ensure it satisfies all constraints. For example, for energy consumption constraints, the verification checks whether the estimated energy consumption under this parameter combination is below a specified threshold; for quality constraints, the verification checks whether the product quality indicators meet the required standards. Each candidate solution undergoes comprehensive verification, and only those satisfying all constraints are marked as feasible and added to the initial feasible solution set.

[0057] Based on an initial set of feasible solutions, a heuristic search algorithm is used to perform a neighborhood search in the decision space to expand the set of feasible solutions. The core idea of ​​neighborhood search is to slightly perturb the control parameters of known feasible solutions and explore new feasible solutions in their surrounding regions. Gaussian perturbation can be used, adding a random value following a normal distribution to each control parameter in the feasible solutions. The perturbation magnitude gradually decreases as the search progresses; larger perturbations are used in the initial stage for broad exploration, while smaller perturbations are used in the later stages for refined searching.

[0058] For new candidate solutions generated through neighborhood search, constraint verification is also performed. The verification process is the same as for the initial candidate solutions, checking whether the newly generated candidate solutions satisfy all constraints. For new candidate solutions that pass verification, they are added to the feasible solution set to further expand the size and diversity of the feasible solution set.

[0059] The neighborhood search process is repeated until the size of the feasible solution set reaches a preset requirement. This preset size can be determined based on the problem complexity and subsequent optimization needs, and is typically set large enough to cover the main feasible regions in the decision space, such as 100 or more feasible solutions. Once the feasible solution set reaches the preset size, it is output as the initial solution set, providing a foundation for subsequent multi-objective optimization processes.

[0060] In practical applications, heuristic search algorithms can include simulated annealing or genetic algorithms. Taking simulated annealing as an example, in addition to the standard neighborhood search mechanism, a temperature parameter is introduced to control the probability of accepting non-improved solutions. When the initial temperature is high, there is a greater probability of accepting non-improved solutions to escape local optima. As the temperature gradually decreases, the algorithm tends to accept only improved solutions, achieving a balance between global exploration and local fine-grained search.

[0061] To improve search efficiency, an adaptive perturbation strategy can be further adopted, dynamically adjusting the perturbation magnitude based on the constraint satisfaction rate. When perturbations in a certain direction frequently lead to constraint violations, the perturbation magnitude in that direction is reduced; when perturbations in a certain direction repeatedly produce feasible solutions, the perturbation in that direction is appropriately increased to accelerate the exploration.

[0062] Through the above steps, feasible solution regions that satisfy the constraints are systematically explored in the decision space, generating a representative and diverse initial solution set, which provides a good foundation for subsequent multi-objective optimization, thereby helping the production system find control parameter configuration schemes that both satisfy the constraints and have good performance.

[0063] In one optional implementation, a gradient optimization algorithm is used to perform local optimization within the feasible solution region, starting from the initial solution set, to solve a multi-device collaborative optimization objective function that aims to minimize inter-device process waiting time and maximize material flow continuity. The optimal control parameter combinations for each production device are obtained, including: The initial solution with the optimal objective function value for multi-device collaborative optimization is selected from the initial solution set as the starting point for gradient optimization; The multi-equipment collaborative optimization objective function is decomposed into a sub-objective function that minimizes the inter-equipment process waiting time and a sub-objective function that maximizes the continuity of material flow. The gradient vectors of the sub-objective functions that minimize the inter-equipment process waiting time and the sub-objective functions that maximize the continuity of material flow are calculated at the starting point to obtain the first sub-objective gradient vector and the second sub-objective gradient vector. Calculate the angle between the gradient vectors of the first sub-target and the second sub-target, and determine the consistency of the optimization directions of the two sub-targets. When the angle is less than a preset angle threshold, the weighted sum of the gradient vectors of the first and second sub-targets is used as the coordinating gradient direction. When the angle is greater than the preset angle threshold, the Pareto front projection method is used to project the gradient vectors of the two targets onto the Pareto improvement direction to obtain the coordinating gradient direction. The control parameters of the starting point are adjusted along the direction of the coordinated gradient. The parameter adjustment step size is determined by a line search method, and the adjusted control parameters are combined as a new starting point. Repeated iterations are performed. When the magnitudes of the gradient vectors of the first and second sub-targets are both less than the preset convergence threshold after multiple consecutive iterations, the combination of control parameters corresponding to the current starting point is taken as the optimal combination of control parameters for each production device.

[0064] When using gradient optimization algorithms for multi-device collaborative optimization, the first step is to select the optimal initial solution from the initial solution set as the starting point. After obtaining the initial solution set, the objective function value for multi-device collaborative optimization corresponding to each initial solution is calculated. This objective function comprehensively considers both minimizing the process waiting time between devices and maximizing the continuity of material flow. The solution with the optimal objective function value is selected as the starting point for gradient optimization to improve the efficiency of subsequent optimization.

[0065] For a selected starting point, the multi-equipment collaborative optimization objective function is decomposed into two sub-objective functions: minimizing inter-equipment process waiting time and maximizing material flow continuity. The inter-equipment process waiting time sub-objective function can be expressed as the sum of time differences between adjacent processes, while the material flow continuity sub-objective function can be quantified by matching the material input and output rates of each piece of equipment. The gradient vectors of these two sub-objective functions are calculated at the starting point, yielding the first sub-objective gradient vector and the second sub-objective gradient vector.

[0066] To coordinate two optimization objectives, it is necessary to determine the consistency of the optimization directions of the two sub-objectives. The degree of consistency is measured by calculating the angle between the two gradient vectors. Specifically, the dot product formula is used: the dot product of the two gradient vectors is divided by the product of their magnitudes, and then the inverse cosine is taken to obtain the angle value. When the calculated angle is less than a preset angle threshold (e.g., 30 degrees), it indicates that the directions of the two optimization objectives are basically consistent, and a weighted sum method can be used to combine the two gradient vectors as the coordinating gradient direction. The weights can be preset according to the relative importance of the two sub-objectives, for example, set to 0.6 and 0.4 respectively.

[0067] When the angle between two gradient vectors exceeds a preset angle threshold, it indicates a significant conflict between the two optimization objectives. In this case, the Pareto front projection method is used to determine the coordinating gradient direction. This method first constructs a plane spanned by the two gradient vectors, then projects possible search directions onto this plane, and selects the direction that can simultaneously improve both objectives or significantly improve the other objective at the expense of one objective as the coordinating gradient direction. In practice, this can be achieved through a linear combination of the two gradient vectors, and the optimal combination coefficients can be searched using a bisection method.

[0068] After determining the direction of the coordinating gradient, the control parameters at the starting point are adjusted along this direction. The adjustment step size is determined using a line search method, such as the Armijo line search rule. This method starts with an initial step size; if the improvement in the objective function value is insufficient at the current step size, the step size is reduced by a certain proportion (e.g., 0.5) until a suitable step size is found. In practice, the initial step size and step size reduction factor are first set, then the objective function value corresponding to the new parameter combination after moving along the direction of the coordinating gradient is calculated. The improvement degree of the old and new objective function values ​​is compared. If the improvement meets the preset conditions, the current step size is accepted; otherwise, the step size is reduced and the attempt is repeated.

[0069] Using the adjusted control parameter combination as a new starting point, the iterative process of gradient calculation, direction coordination, and parameter adjustment is repeated. In each iteration, the magnitudes of the gradient vectors of the first and second sub-targets are monitored. When, after several consecutive iterations (e.g., three consecutive iterations), the magnitudes of both gradient vectors are less than a preset convergence threshold (e.g., 0.001), the optimization process can be considered to have converged to a local optimum. At this point, the current parameter combination is taken as the optimal control parameter combination for each production device.

[0070] In one optional implementation, the Pareto front projection method is used to project the two target gradient vectors onto the Pareto improvement direction to obtain the coordinated gradient direction, including: A Pareto improvement cone is constructed, which is defined by the set of all search directions that simultaneously decrease the sub-objective function values ​​corresponding to the gradient vectors of the first and second sub-objectives. The coordinated gradient direction is determined by solving the optimal projection direction of the weighted combination vector of the first sub-objective gradient vector and the second sub-objective gradient vector within the Pareto improved cone. The optimal projection direction is obtained by maximizing the magnitude of the projection vector within the Pareto improved cone while simultaneously satisfying the constraint condition that both sub-objective function values ​​decrease. Calculate the dot product of the coordinated gradient direction and the gradient vector of the first sub-target, and calculate the dot product of the coordinated gradient direction and the gradient vector of the second sub-target to verify that the coordinated gradient direction simultaneously satisfies the descent requirements of both sub-targets; The verified coordinated gradient direction will be used as the search direction for subsequent parameter adjustments.

[0071] The process of projecting two target gradient vectors onto the Pareto improvement direction using the Pareto front projection method to obtain the reconciled gradient direction is as follows: First, we construct the Pareto improvement cone. The Pareto improvement cone is the set of all search directions that simultaneously decrease the sub-objective function values ​​corresponding to the gradient vectors of the first and second sub-objectives. Assuming the gradient vector of the first sub-objective is g1 and the gradient vector of the second sub-objective is g2, the Pareto improvement cone can be defined as the set of vectors d that satisfy the following condition: (g1) T d < 0 and (g2) T d < 0. This means that as you move along direction d, the values ​​of both objective functions will decrease, achieving a Pareto improvement in multi-objective optimization.

[0072] Next, the coordinated gradient direction is determined by solving for the optimal projection direction of the weighted combination vector of the first and second sub-objective gradient vectors within the Pareto improvement cone. Let the weighted combination of the first and second sub-objective gradient vectors be g = λ1g1 + λ2g2, where λ1 and λ2 are weight coefficients, and λ1 + λ2 = 1, λ1≥0, λ2≥0. The optimal projection direction D needs to satisfy the following: within the Pareto improvement cone, D maximizes the magnitude of the projection vector, while simultaneously satisfying the constraint that the values ​​of both sub-objective functions decrease.

[0073] This optimal projection problem can be transformed into the following optimization problem: maximize ‖d‖ while satisfying (g1) T d < 0 and (g2) T The constraints are d < 0 and ||d|| = 1. The modulus is limited to a unit length for ease of solution. Solving this optimization problem yields a unit vector, which represents the direction of the desired coordinating gradient. The actual coordinating gradient direction can be obtained by multiplying by an appropriate step size factor.

[0074] In practical calculations, this projection problem can be solved using a geometric method. First, calculate the angle θ between g1 and g2. If θ < 180 degrees, the two gradient vectors are not completely opposite, and there exists a feasible Pareto improvement direction. The optimal projection direction D lies along the angle bisector between -g1 and -g2. If g1 and g2 are collinear and in opposite directions, any direction perpendicular to them can be chosen as D. If g1 and g2 are collinear and in the same direction, a direction opposite to them is chosen as D.

[0075] After calculating the coordinating gradient direction D, it is necessary to verify that it simultaneously satisfies the descent requirements of both sub-objectives. The specific verification method is to calculate the inner product of D and g1, and the inner product of D and g2. If both inner products are negative, it indicates that the function values ​​of both sub-objectives decrease when moving along the D direction, and the verification passes.

[0076] In a specific application scenario, suppose the first sub-objective is the training error of the neural network model, and the second sub-objective is the model complexity. The gradient vector g1 of the first sub-objective points in the direction of the fastest increase in training error, and the gradient vector g2 of the second sub-objective points in the direction of the fastest increase in model complexity. By solving for the coordinating gradient direction D, an optimal direction that simultaneously reduces training error and model complexity can be found, thus achieving a balance between model performance and simplicity.

[0077] Finally, the validated coordinating gradient direction D is used as the search direction for subsequent parameter adjustments. During parameter updates, new parameter values ​​can be calculated by subtracting the coordinating gradient direction from the current parameter value and multiplying by an appropriate step size, and can be determined using methods such as line search.

[0078] This Pareto front projection method effectively coordinates multiple conflicting optimization objectives, finding a balanced search direction and avoiding the problem of mutually canceling objective functions that may occur in traditional multi-objective optimization. This improves optimization efficiency and convergence speed. This method is particularly suitable for optimization problems involving multi-objective balance, such as deep learning and control system tuning.

[0079] In an optional implementation, the method further includes constructing a sub-objective function for minimizing inter-equipment process waiting time and a sub-objective function for maximizing material flow continuity: Traverse each directed edge in the dynamic cooperative topology and identify the source device node and target device node corresponding to each directed edge; The material output completion time is determined based on the source device node, and the material receiving readiness time is determined based on the target device node. The time difference between the material receiving readiness time and the material output completion time is used as the inter-device waiting time corresponding to the directed edge. The waiting times between devices corresponding to all directed edges in the dynamic cooperative topology are summed to obtain the total waiting time of the overall inter-device process waiting loss. The total waiting time is used as the optimization objective of the sub-objective function for minimizing the inter-device process waiting time. Traverse each directed edge in the dynamic cooperative topology, determine the material transfer timing conditions from the material transfer constraints marked on each directed edge, and determine whether the material output time of the source device node and the material receiving time of the target device node satisfy the material transfer timing conditions; The number of directed edges that satisfy the material transfer timing conditions is counted, and the ratio of the number of directed edges that satisfy the material transfer timing conditions to the total number of directed edges is calculated as the material transfer timing compliance rate. Traverse each device node in the dynamic collaborative topology, extract the material batch sequence processed by each device node, and count the number of process state switching between adjacent material batches as the process switching count of the device node; The total number of process switching times is obtained by summing the number of process switching times of all device nodes in the dynamic collaborative topology. The material transfer timing compliance rate is weighted and summed with the reciprocal of the total number of process switching times to obtain a flow continuity index that characterizes the continuity of material flow. The flow continuity index is used as the optimization objective of the sub-objective function for maximizing the continuity of material flow.

[0080] Minimizing process waiting time between equipment and maximizing the continuity of material flow are key factors in achieving efficient production.

[0081] The directed edges in the dynamic collaborative topology are traversed to identify the source and target device nodes for each directed edge. For example, directed edge E1 represents the material transfer relationship from device S1 to device S2. The source device node of E1 is determined to be S1, and the target device node is determined to be S2, based on the node identifier.

[0082] Determine the waiting time between devices. Based on the process parameters of the source device node and the processing status of the current material batch, calculate the material output completion time To. For example, if device S1 processes material batch B1 at 9:00 AM and the processing time is 45 minutes, then To is 9:45 AM. Simultaneously, determine the material receiving readiness time Tr based on the status of the target device node. If device S2 is currently processing other batches and is expected to receive new material at 10:00 AM, then Tr is 10:00 AM. Calculate the waiting time from S1 to S2: Tw = Tr - To = 15 minutes, representing the idle waiting time of the material between devices.

[0083] The total system waiting time, Ttotal, is obtained by summing the waiting times of all directed edges. Assuming there are 5 directed edges in the system with corresponding waiting times of 15 minutes, 8 minutes, 12 minutes, 20 minutes, and 5 minutes, then Ttotal = 60 minutes. Ttotal is used as the optimization objective of the sub-objective function for minimizing inter-device process waiting time; the optimization objective is to minimize the value of Ttotal.

[0084] In assessing the continuity of material flow, each directed edge is first traversed to check whether its material transfer timing conditions are met. For example, for temperature-sensitive material transfer, there may be a timing constraint that "material transfer must be completed within 10 minutes after output." If source device S1 outputs material at 9:45 and target device S2 receives it at 9:52, satisfying the 10-minute timing constraint, then the directed edge meets the material transfer timing conditions.

[0085] Count the number of directed edges Nc that meet the timing conditions, and calculate the ratio of this ratio to the total number of directed edges Nt as the material transfer timing compliance rate R = Nc / Nt. If there are 20 directed edges in the system, and 18 of them meet the timing conditions, then R = 18 / 20 = 90%.

[0086] Then, the batch sequence of materials processed by each equipment node is extracted, and the changes in process status between adjacent batches are analyzed. For example, if equipment S3 continuously processes three batches B1, B2, and B3, where B1 and B2 use the same process parameters, while B3 requires adjustments to temperature and pressure, then the process switching count for S3 is 1. Similarly, the process switching counts for all equipment nodes are counted and summed to obtain the total process switching count Cs.

[0087] The material flow continuity index Fc is calculated by weighting the material transfer timing compliance rate R and the reciprocal of the total number of process changeovers Cs: Fc = α·R + β·(1 / Cs), where α and β are weighting coefficients determined according to actual production needs. For example, if α=0.7, β=0.3, R=90%, and Cs=8, then Fc=0.7×0.9+0.3×(1 / 8)=0.6675. The higher the value of this index, the better the material flow continuity, and it serves as the optimization objective of the sub-objective function for maximizing material flow continuity.

[0088] In practical applications, these two sub-objective functions can be combined into the overall optimization objective to form a multi-objective optimization problem. For example, a weighted summation method can be used: F = w1·(1 / Ttotal) + w2·Fc, where w1 and w2 are weight coefficients that are adjusted according to the production focus. By solving this optimization objective, a dynamic collaborative scheduling scheme can be obtained that reduces waiting time between equipment and improves the continuity of material flow.

[0089] This method is particularly suitable for scenarios involving multiple devices working together in mass production. By constructing a reasonable sub-objective function, it can significantly improve production efficiency, reduce energy consumption and material waste, and at the same time ensure the stability of product quality.

[0090] A second aspect of the present invention provides a dual-extract oil essence capsule, comprising: The whitening capsule formula and the anti-aging capsule formula are prepared in parts by weight, and the whitening capsule formula and the anti-aging capsule formula are filled into the same capsule in the same volume ratio; The whitening formula comprises the following components and parts by weight: 30-40 parts of coconut oil alcohol-caprylate / capric acid ester, 32-36 parts of isododecane, 11-18 parts of isohexadecane, 2-5 parts of hydrogenated styrene / isoprene copolymer, 3.5-5.5 parts of pentaerythritol tetrabis-tert-butylhydroxyhydrogenated cinnamic acid ester, 2.5-4.5 parts of oil-soluble niacinamide-containing whitening active ingredient combination, and 0 parts of phenylethyl resorcinol-containing whitening active ingredient combination. 5-1.5 parts, a whitening active ingredient combination containing tranexamic acid and nonapeptide-1 (0.2-0.6 parts), bisabolol (0.2-0.6 parts), citrus essential oil blend (0.15-0.5 parts), myristyl nicotinate (0.05-0.25 parts), palm bark extract (0.08-0.12 parts), glycyrrhizin (0.08-0.12 parts), stearyl glycyrrhetinic acid ester (0.04-0.12 parts), and tocopherol (0.04-0.12 parts); The anti-aging capsule formulation comprises the following components and parts by weight: 30-45 parts of cyclopentamethoxysiloxane, 24-40 parts of polydimethylsiloxane, 3-6.5 parts of hexyldecyl alcohol, 3-6 parts of octyl polymethylsiloxane, 2.5-4.5 parts of divinyl polydimethylsiloxane / polydimethylsiloxane crosspolymer, 2.5-4.5 parts of polydimethylsiloxane / vinyl polydimethylsiloxane crosspolymer, 2-3 parts of coconut oil alcohol-caprylate / caprylate, and 0.9 parts of phytosterol / octyldodecyl lauroyl glutamate. -2.2 parts, 0.4-1.2 parts of anti-aging active ingredient combination containing hybrid rose flower extract, 0.2-0.6 parts of anti-aging active ingredient combination containing sodium DNA and ergothioneine, 0.3-0.5 parts of citrus essential oil compound, 0.15-0.5 parts of bisabolol combination containing ginger root extract, 0.04-0.12 parts of rosewood bark extract, 0.04-0.12 parts of stearyl glycyrrhetinic acid ester, 0.04-0.12 parts of tocopherol, and 0.0001-0.001 parts of sugar isomer combination.

[0091] It should also be noted that the dual-extract skincare capsules of this application utilize a precision plant-based encapsulation system to achieve independent encapsulation and synergistic release of the whitening and anti-aging formulas. The plant-based shell process uses modified starch, gelatin substitutes, and cellulose derivatives as the main components of the capsule shell. Modified starch provides the capsule's mechanical strength, gelatin substitutes enhance the shell's flexibility, and cellulose derivatives regulate the shell's dissolution rate. The shell material is melted at high temperatures and then cast using a precision mold. The mold temperature is controlled between 85 and 95 degrees Celsius, the casting pressure is maintained at 0.3 to 0.5 MPa, and the cooling time is set to 8 to 12 seconds. The formed capsule shell exhibits a dual-cavity structure, with the two independent chambers physically separated by a flexible diaphragm. The diaphragm thickness is 0.08 to 0.12 mm, and the diaphragm material is a biodegradable plant fiber composite membrane.

[0092] The dual-phase freshness-locking technology maintains the stability of active ingredients by separately nitrogen-filling the whitening and anti-aging formulations. The whitening formulation undergoes deoxygenation treatment before filling, reducing the oxygen content to below 0.5% by volume. After material injection in the whitening chamber, the filling and packaging equipment immediately introduces high-purity nitrogen for headspace replacement. The nitrogen purity is no less than 99.99%, and the replacement time lasts 3 to 5 seconds, reducing the oxygen content in the chamber to below 0.2% by volume. The anti-aging formulation also undergoes nitrogen-filling protection, with the nitrogen pressure controlled at 0.05 to 0.08 MPa to ensure uniform nitrogen distribution on the material surface, forming a protective layer. After both chambers are filled with nitrogen, the filling head moves to the sealing position, and the capsule cap is joined to the capsule body using ultrasonic welding technology. The welding power is set to 80 to 120 watts, the welding time is 1.5 to 2.5 seconds, and the welding temperature is controlled at 70 to 80 degrees Celsius. The heat generated during welding causes the plant material at the contact surface between the capsule cap and the capsule body to melt and bond together. After cooling, a sealed structure is formed with a sealing strength of 0.8 to 1.2 Newtons and a leakage rate of less than 0.01%.

[0093] A third aspect of the present invention provides a digital process control system for a dual-extract skincare capsule production line, comprising: The first unit is used to acquire real-time operating status data of multiple production equipment in the dual-extract skincare capsule production line, parse the real-time operating status data, and extract the current process stage identifier and corresponding process constraints of each production equipment; The second unit is used to construct a dynamic collaborative topology structure that reflects the material transfer relationship and process connection relationship between equipment based on the process stage identifier. In the dynamic collaborative topology structure, the state transition conditions of each equipment node and the material transfer constraints of the edge connection are marked. Based on the topological connectivity and node state distribution of the dynamic collaborative topology structure, the bidirectional collaborative constraint relationship between each production equipment is determined. The third unit is used to transform the bidirectional collaborative constraint relationship into a set of constraints for a multi-device collaborative optimization problem, and to transform the process constraints corresponding to each production device into the boundary values ​​of decision variables; The fourth unit is used to explore feasible solution regions in the decision space that satisfy the set of constraints using a heuristic search algorithm and generate an initial solution set. Then, using a gradient optimization algorithm, it performs local optimization within the feasible solution region starting from the initial solution set to solve the multi-device collaborative optimization objective function with the optimization objectives of minimizing the process waiting time between equipment and maximizing the continuity of material flow, and obtains the optimal combination of control parameters for each production equipment.

[0094] Fourth aspect of the embodiments of the present invention, A computer-readable storage medium is provided, having stored thereon computer program instructions that, when executed by a processor, implement the aforementioned method.

[0095] This invention can be a method, apparatus, system, and / or computer program product. The computer program product may include a computer-readable storage medium having computer-readable program instructions loaded thereon for performing various aspects of the invention.

[0096] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A digital process control method for a double-extract skincare capsule production line, characterized in that, include: The system acquires real-time operating status data of multiple production devices in the dual-extract skincare capsule production line, parses the real-time operating status data, and extracts the current process stage identifier and corresponding process constraints of each production device. Based on the process stage identifiers, a dynamic collaborative topology is constructed that reflects the material transfer relationship and process connection relationship between equipment. The state transition conditions of each equipment node and the material transfer constraints of the edge connections are marked in the dynamic collaborative topology. Based on the topological connectivity and node state distribution of the dynamic collaborative topology, the bidirectional collaborative constraint relationship between each production equipment is determined. The bidirectional collaborative constraint relationship is transformed into a set of constraints for a multi-equipment collaborative optimization problem, and the process constraints corresponding to each production equipment are transformed into the boundary values ​​of decision variables. A heuristic search algorithm is used to explore feasible solution regions in the decision space that satisfy the set of constraints and generate an initial solution set. A gradient optimization algorithm is then used to perform local optimization within the feasible solution region, starting from the initial solution set, to solve the multi-equipment collaborative optimization objective function with the optimization objectives of minimizing the process waiting time between equipment and maximizing the continuity of material flow, thereby obtaining the optimal combination of control parameters for each production equipment.

2. The method according to claim 1, characterized in that, Based on the process stage identifiers, a dynamic collaborative topology is constructed that reflects the material transfer relationships and process connection relationships between equipment. The state transition conditions of each equipment node and the material transfer constraints of the edge connections are marked in the dynamic collaborative topology, including: Each production device is mapped to a device node in the dynamic collaborative topology, and the material transfer path between devices is mapped to a directed edge connecting the device nodes; According to the process stage identifier of each production equipment, the current process state is marked on the corresponding equipment node. According to the process flow rules, the state transition conditions are marked on each equipment node. The state transition conditions define the prerequisites that the equipment node must meet to transition from the current process state to the next process state. Material transfer constraints are marked on each directed edge. The material transfer constraints define the material state requirements and transfer timing requirements that must be met when the source device node transfers materials to the target device node.

3. The method according to claim 2, characterized in that, Based on the topological connectivity and node state distribution of the dynamic collaborative topology, the bidirectional collaborative constraint relationships between various production devices are determined, including: Based on the topological connectivity of the dynamic collaborative topology, each directed edge in the dynamic collaborative topology is traversed, and the source device node and target device node corresponding to each directed edge are identified; the state transition constraints related to material output are extracted from the state transition conditions marked by the source device node, and the material transfer constraints are extracted from the directed edges connecting the source device node and the target device node; The state transition constraints and the material transfer constraints are logically conjuncted to obtain the forward coordination constraints between the source device node and the target device node; the state transition constraints related to material receiving are extracted from the state transition conditions marked by the target device node, and these state transition constraints are used as the reverse coordination constraints of the target device node on the source device node; The forward collaborative constraint and the reverse collaborative constraint are combined to form a bidirectional collaborative constraint relationship.

4. The method according to claim 1, characterized in that, The heuristic search algorithm is used to explore feasible solution regions in the decision space that satisfy the set of constraints and generate an initial solution set, including: The dimensions and boundary range of the decision space are defined based on the set of constraints. Multiple candidate solutions are randomly generated in the decision space, and each candidate solution corresponds to a set of control parameter values ​​for each production equipment. Each candidate solution is subjected to constraint verification to determine whether it satisfies all constraints in the constraint set. Candidate solutions that satisfy all constraints are marked as feasible solutions. A heuristic search algorithm is used to perform a neighborhood search in the decision space based on the feasible solution. New candidate solutions are generated by perturbing the control parameters of the feasible solution, and the constraints of the new candidate solutions are verified. The new candidate solutions that pass the verification are added to the feasible solution set. The neighborhood search process is repeated until the size of the feasible solution set reaches the preset size requirement. The feasible solution set is then used as the initial solution set.

5. The method according to claim 1, characterized in that, Using the gradient optimization algorithm, starting from the initial solution set, local optimization is performed within the feasible solution region to solve the multi-equipment collaborative optimization objective function, which aims to minimize the process waiting time between equipment and maximize the continuity of material flow. The optimal control parameter combinations for each production equipment are obtained, including: The initial solution with the optimal objective function value for multi-device collaborative optimization is selected from the initial solution set as the starting point for gradient optimization; The multi-equipment collaborative optimization objective function is decomposed into a sub-objective function that minimizes the inter-equipment process waiting time and a sub-objective function that maximizes the continuity of material flow. The gradient vectors of the sub-objective functions that minimize the inter-equipment process waiting time and the sub-objective functions that maximize the continuity of material flow are calculated at the starting point to obtain the first sub-objective gradient vector and the second sub-objective gradient vector. Calculate the angle between the gradient vectors of the first sub-target and the second sub-target, and determine the consistency of the optimization directions of the two sub-targets. When the angle is less than a preset angle threshold, the weighted sum of the gradient vectors of the first and second sub-targets is used as the coordinating gradient direction. When the angle is greater than the preset angle threshold, the Pareto front projection method is used to project the gradient vectors of the two targets onto the Pareto improvement direction to obtain the coordinating gradient direction. The control parameters of the starting point are adjusted along the direction of the coordinated gradient. The parameter adjustment step size is determined by a line search method, and the adjusted control parameters are combined as a new starting point. Repeated iterations are performed. When the magnitudes of the gradient vectors of the first and second sub-targets are both less than the preset convergence threshold after multiple consecutive iterations, the combination of control parameters corresponding to the current starting point is taken as the optimal combination of control parameters for each production device.

6. The method according to claim 5, characterized in that, The Pareto front projection method is used to project the two target gradient vectors onto the Pareto improvement direction to obtain the coordinated gradient direction, which includes: A Pareto improvement cone is constructed, which is defined by the set of all search directions that simultaneously decrease the sub-objective function values ​​corresponding to the gradient vectors of the first and second sub-objectives. The coordinated gradient direction is determined by solving the optimal projection direction of the weighted combination vector of the first sub-objective gradient vector and the second sub-objective gradient vector within the Pareto improved cone. The optimal projection direction is obtained by maximizing the magnitude of the projection vector within the Pareto improved cone while simultaneously satisfying the constraint condition that both sub-objective function values ​​decrease. Calculate the dot product of the coordinated gradient direction and the gradient vector of the first sub-target, and calculate the dot product of the coordinated gradient direction and the gradient vector of the second sub-target to verify that the coordinated gradient direction simultaneously satisfies the descent requirements of both sub-targets; The verified coordinated gradient direction will be used as the search direction for subsequent parameter adjustments.

7. The method according to claim 5, characterized in that, The method also includes constructing a sub-objective function that minimizes process waiting time between equipment and a sub-objective function that maximizes the continuity of material flow: Traverse each directed edge in the dynamic cooperative topology and identify the source device node and target device node corresponding to each directed edge; The material output completion time is determined based on the source device node, and the material receiving readiness time is determined based on the target device node. The time difference between the material receiving readiness time and the material output completion time is used as the inter-device waiting time corresponding to the directed edge. The waiting times between devices corresponding to all directed edges in the dynamic cooperative topology are summed to obtain the total waiting time of the overall inter-device process waiting loss. The total waiting time is used as the optimization objective of the sub-objective function for minimizing the inter-device process waiting time. Traverse each directed edge in the dynamic cooperative topology, determine the material transfer timing conditions from the material transfer constraints marked on each directed edge, and determine whether the material output time of the source device node and the material receiving time of the target device node satisfy the material transfer timing conditions; The number of directed edges that satisfy the material transfer timing conditions is counted, and the ratio of the number of directed edges that satisfy the material transfer timing conditions to the total number of directed edges is calculated as the material transfer timing compliance rate. Traverse each device node in the dynamic collaborative topology, extract the material batch sequence processed by each device node, and count the number of process state switching between adjacent material batches as the process switching count of the device node; The total number of process switching times is obtained by summing the number of process switching times of all device nodes in the dynamic collaborative topology. The material transfer timing compliance rate is weighted and summed with the reciprocal of the total number of process switching times to obtain a flow continuity index that characterizes the continuity of material flow. The flow continuity index is used as the optimization objective of the sub-objective function for maximizing the continuity of material flow.

8. A double-extract oil essence capsule produced using any one of claims 7, characterized in that, include: The whitening capsule formula and the anti-aging capsule formula are prepared in parts by weight, and the whitening capsule formula and the anti-aging capsule formula are filled into the same capsule in the same volume ratio; The whitening formula comprises the following components and parts by weight: 30-40 parts of coconut oil alcohol-caprylate / capric acid ester, 32-36 parts of isododecane, 11-18 parts of isohexadecane, 2-5 parts of hydrogenated styrene / isoprene copolymer, 3.5-5.5 parts of pentaerythritol tetrabis-tert-butylhydroxyhydrogenated cinnamic acid ester, 2.5-4.5 parts of oil-soluble niacinamide-containing whitening active ingredient combination, and 0 parts of phenylethyl resorcinol-containing whitening active ingredient combination. 5-1.5 parts, a whitening active ingredient combination containing tranexamic acid and nonapeptide-1 (0.2-0.6 parts), bisabolol (0.2-0.6 parts), citrus essential oil blend (0.15-0.5 parts), myristyl nicotinate (0.05-0.25 parts), palm bark extract (0.08-0.12 parts), glycyrrhizin (0.08-0.12 parts), stearyl glycyrrhetinic acid ester (0.04-0.12 parts), and tocopherol (0.04-0.12 parts); The anti-aging capsule formulation comprises the following components and parts by weight: 30-45 parts of cyclopentamethoxysiloxane, 24-40 parts of polydimethylsiloxane, 3-6.5 parts of hexyldecyl alcohol, 3-6 parts of octyl polymethylsiloxane, 2.5-4.5 parts of divinyl polydimethylsiloxane / polydimethylsiloxane crosspolymer, 2.5-4.5 parts of polydimethylsiloxane / vinyl polydimethylsiloxane crosspolymer, 2-3 parts of coconut oil alcohol-caprylate / caprylate, and 0.9 parts of phytosterol / octyldodecyl lauroyl glutamate. -2.2 parts, 0.4-1.2 parts of anti-aging active ingredient combination containing hybrid rose flower extract, 0.2-0.6 parts of anti-aging active ingredient combination containing sodium DNA and ergothioneine, 0.3-0.5 parts of citrus essential oil compound, 0.15-0.5 parts of bisabolol combination containing ginger root extract, 0.04-0.12 parts of rosewood bark extract, 0.04-0.12 parts of stearyl glycyrrhetinic acid ester, 0.04-0.12 parts of tocopherol, and 0.0001-0.001 parts of sugar isomer combination.

9. A digital process control system for a dual-extract skincare capsule production line, used to implement the method as described in any one of claims 1-7, characterized in that, include: The first unit is used to acquire real-time operating status data of multiple production equipment in the dual-extract skincare capsule production line, parse the real-time operating status data, and extract the current process stage identifier and corresponding process constraints of each production equipment; The second unit is used to construct a dynamic collaborative topology structure that reflects the material transfer relationship and process connection relationship between equipment based on the process stage identifier. In the dynamic collaborative topology structure, the state transition conditions of each equipment node and the material transfer constraints of the edge connection are marked. Based on the topological connectivity and node state distribution of the dynamic collaborative topology structure, the bidirectional collaborative constraint relationship between each production equipment is determined. The third unit is used to transform the bidirectional collaborative constraint relationship into a set of constraints for a multi-device collaborative optimization problem, and to transform the process constraints corresponding to each production device into the boundary values ​​of decision variables; The fourth unit is used to explore feasible solution regions in the decision space that satisfy the set of constraints using a heuristic search algorithm and generate an initial solution set. Then, using a gradient optimization algorithm, it performs local optimization within the feasible solution region starting from the initial solution set to solve the multi-device collaborative optimization objective function with the optimization objectives of minimizing the process waiting time between equipment and maximizing the continuity of material flow, and obtains the optimal combination of control parameters for each production equipment.

10. A computer-readable storage medium having computer program instructions stored thereon, characterized in that, When the computer program instructions are executed by the processor, they implement the method described in any one of claims 1 to 7.

Images