CN-121998394-A - Intelligent clothing production ranking method and system based on Internet of things

CN121998394ACN 121998394 ACN121998394 ACN 121998394ACN-121998394-A

Abstract

The invention belongs to the technical field of industrial Internet of things and intelligent manufacturing, and particularly relates to an intelligent ranking method and system for clothing production based on the Internet of things. The method comprises the steps of retrieving matched candidate genotypes from a production mode genetic map library and analyzing the candidate genotypes into an initial ranking scheme, calculating personnel-equipment-process collaborative fitness coefficients of each station in the scheme, constructing an optimized objective function integrating collaborative fitness, historical environment fitness and production constraint, solving and outputting the optimal ranking scheme by adopting an improved genetic algorithm, dynamically constructing a process flow potential energy field in production execution, and generating and executing task shunting scheduling instructions according to potential energy gradients. The invention realizes the full-flow intelligent ranking from the multiplexing of historical experience, global multi-objective optimization to dynamic real-time scheduling, effectively solves the technical problems of depending experience, single optimization objective and insufficient dynamic response of scheduling in the flexible production line of clothing, and remarkably improves the scheduling efficiency, the balance rate of the production line and the self-adaptive capacity of the system.

Inventors

Miao Zishi
LI XIAOXU
ZHANG JINHAO
LI ZHIPENG

Assignees

北京宜通华瑞科技有限公司

Dates

Publication Date: 20260508
Application Date: 20260112

Claims (10)

1. An intelligent clothing production ranking method based on the Internet of things is characterized by comprising the following steps: Step S1, retrieving at least one candidate genotype matched with the current order process characteristics and the production line resource state from a production mode genetic map library, wherein the production mode genetic map library stores genotype codes generated based on a historical successful ranking scheme, and the genotype codes at least comprise personnel skill feature vectors, equipment function feature vectors and logic relationship topology among working procedures; S2, analyzing the searched candidate genotypes into an initial ranking scheme, and calculating a collaborative fitness coefficient of a personnel-equipment-procedure combination distributed by each station in the initial ranking scheme, wherein the collaborative fitness coefficient is obtained by fusion based on a personnel skill quantization value, an equipment capacity quantization value and a procedure demand quantization value; step S3, constructing an optimized objective function which takes the overall collaborative fitness coefficient as a core, considers the historical environment fitness related to the candidate genotypes and meets the constraint of the intersection period and the load, and adopts a genetic algorithm to solve and output and execute an optimal ranking scheme based on the initial ranking scheme; And S4, dynamically constructing a process flow potential energy field according to the real-time task load and the processing capacity of each station in the production execution, identifying a bottleneck station in the process flow potential energy field, generating a task shunting scheduling instruction according to the potential energy gradient from the bottleneck station to the adjacent station, and executing the task shunting scheduling instruction.
2. The intelligent clothes production ranking method based on the internet of things according to claim 1, wherein step S1 adopts a dynamic evolution and multilayer progressive search mechanism, and comprises the following steps: When the new ranking scheme reaches a preset efficacy standard through production verification, automatically extracting personnel skill feature vectors, equipment function feature vectors, logic relationship topology among working procedures and scheduling decision sequence features of the new ranking scheme, correlating the logic relationship topology with execution environment and efficacy data to form new genotype codes, and incrementally merging the new genotype codes into a production mode genetic map library in a mode of minimizing topology conflict; Coarse-grained screening is carried out based on the topological matching degree of the logical relationship among the working procedures, and a primary candidate genotype set is obtained; in the primary candidate genotype set, combining the compatibility of the personnel skill feature vector and the equipment function feature vector to carry out fine granularity matching to obtain a secondary candidate genotype set; And in the secondary candidate genotype set, performing efficiency sequencing according to the reproducibility of the historical scheduling decision sequence under the current resource constraint, and outputting candidate genotypes matched with the current order process characteristics and the production line resource state.
3. The intelligent ranking method for clothing production based on the internet of things according to claim 1, wherein step S2 adopts a state-aware dynamic skill evolution and fusion method, and comprises the following steps: mapping the personnel skill feature vector, the equipment function feature vector and the logic relationship topology among the working procedures contained in the candidate genotype into specific station, personnel, equipment and working procedure distribution relations to generate an initial ranking scheme; Constructing a process proficiency updating algorithm comprising a time decay function and a reinforcement learning function based on personnel allocation in the initial ranking scheme; the process proficiency is positively updated according to the task completion data of the operator and the process proficiency updating algorithm, and is attenuated according to the non-operation period, so that the updated process proficiency is obtained; Calculating a personnel skill quantification value based on the updated process proficiency; Based on equipment allocation and equipment internet of things data in the initial ranking scheme, calculating a performance reduction coefficient reflecting the current running health state, and calculating an equipment capacity quantization value by combining equipment function compatibility; And dynamically adjusting the weight of the personnel skill quantized value and the equipment capacity quantized value in the fusion calculation according to the operator fatigue degree data and the performance reduction coefficient acquired in real time, and then fusing the personnel skill quantized value and the equipment capacity quantized value with the process demand quantized value to generate a collaborative fitness coefficient.
4. The intelligent clothes production ranking method based on the Internet of things according to claim 3, wherein the calculation of the process requirement quantized value is generated through a multidimensional weighting model according to the process standard working hour, the process difficulty coefficient and the accuracy and timeliness double requirements of a current order on a process based on process distribution in an initial ranking scheme.
5. The intelligent clothes production ranking method based on the internet of things according to claim 1, wherein the step S3 adopts an environment fitness prediction and parameter self-adaptive optimization method based on reinforcement learning, and the method comprises the following steps: Defining an optimization objective function as F=w1×C+w2×E+w3×D, wherein C represents an overall collaborative fitness coefficient, E represents a historical environmental fitness predicted value, D represents an intersection urgency factor, w1, w2 and w3 are dynamic weight coefficients, w1+w2+w3=1, and the optimization objective function solves preset constraint conditions which need to meet station load rate, material circulation distance and production change time; constructing a production environment feature space containing order attributes, personnel states, equipment working conditions and material supply; Calculating the multidimensional fuzzy similarity of the current production environment and the genotype historical environment in the production environment feature space, and carrying out weighted fusion on the historical efficiency data by taking the multidimensional fuzzy similarity as a weight to obtain a historical environment adaptability predicted value E; Establishing a reinforcement learning environment taking actual comprehensive production benefits of a ranking scheme as rewards, taking weight coefficients w1, w2 and w3 in an optimization objective function F as adjustable actions, outputting a weight dynamic adjustment strategy for maximizing long-term rewards through strategy learning, and adjusting values of w1, w2 and w 3; And constructing an initial population based on the initial ranking scheme, applying an optimized objective function F of the adjusted weight coefficient, executing the iterative processes of selection, intersection and variation of the genetic algorithm until the convergence condition is met, outputting an optimal ranking scheme and driving a production system to execute.
6. The intelligent ranking method for clothing production based on the internet of things according to claim 5, wherein the genetic algorithm adopts an adaptive improvement mechanism based on population diversity feedback, and the method comprises the following steps: when the initial population is generated, screening is carried out according to the collaborative adaptation coefficient, and only a scheme that the collaborative adaptation coefficient is higher than a preset threshold value is reserved to form the initial population; In the iteration process, dynamically adjusting the crossing rate and the variation rate according to constraint satisfaction degree of the current generation optimal individuals and population gene diversity indexes; and adopting elite reservation and self-adaption generation gap strategy to maintain global searching capability while guaranteeing convergence speed.
7. The intelligent clothes production ranking method based on the internet of things according to claim 1, wherein the potential energy field dynamic scheduling step adopts a potential energy propagation model and a historical efficiency path optimizing method, and the method comprises the following steps: Defining potential energy propagation rules based on a process topology network; The method comprises the steps of calculating the primary potential energy of each station due to self task load and processing capacity, calculating additional potential energy generated by upstream primary potential energy to downstream according to a potential energy transmission rule, and superposing the primary potential energy and the additional potential energy to form a global process flow potential energy field reflecting systematic blocking risks; And calculating potential energy gradients from the bottleneck stations to each adjacent station, predicting secondary bottleneck risks possibly caused by shunting by combining historical shunting success rate and average efficiency hoisting data of each path, performing multi-objective comprehensive scoring, selecting an optimal path to generate a task shunting scheduling instruction, and executing the task shunting scheduling instruction.
8. The intelligent clothes production ranking method based on the Internet of things according to claim 7, wherein execution of task diversion scheduling instructions requires real-time process segmentability verification and target station equipment compatibility verification, and feasibility of scheduling actions is ensured; after the scheduling is executed, the change of the process flow potential energy field is monitored in real time, and if new unbalance is caused, local rebalancing or global rebalancing is triggered.
9. The internet of things-based garment production intelligent ranking method of claim 1, further comprising: Recording the whole process data from generation, optimization and execution to dynamic scheduling of each ranking scheme and the final production efficiency index; Based on the whole process data, updating the efficiency data of genotype codes in a production mode genetic map library, calibrating parameters in a collaborative fitness coefficient calculation model, performing reinforcement learning on strategies in an optimization algorithm, and correcting propagation rules in a process flow potential energy field model.
10. Clothing production intelligence ranking system based on thing networking, its characterized in that includes: The system comprises a history mode searching module, a production mode genetic map library, a processing module and a processing module, wherein the history mode searching module is used for searching at least one candidate genotype matched with the current order process characteristics and the production line resource state from the production mode genetic map library, the production mode genetic map library stores genotype codes generated based on a history successful ranking scheme, and the genotype codes at least comprise personnel skill characteristic vectors, equipment function characteristic vectors and logic relationship topology among working procedures; The adaptation degree quantization and scheme initialization module is used for analyzing the searched candidate genotypes into an initial ranking scheme and calculating a collaborative adaptation degree coefficient of a personnel-equipment-procedure combination distributed by each station in the initial ranking scheme, wherein the collaborative adaptation degree coefficient is obtained by fusion based on a personnel skill quantization value, an equipment capacity quantization value and a procedure demand quantization value; The multi-objective optimization module is used for constructing an optimization objective function which takes the overall collaborative fitness coefficient as a core, considers the historical environment fitness related to the candidate genotypes and meets the constraint of the intersection period and the load, adopts a genetic algorithm to solve based on an initial ranking scheme, and outputs and executes an optimal ranking scheme; And the potential energy field dynamic scheduling module is used for dynamically constructing a process flow potential energy field according to the real-time task load and the processing capacity of each station in the production execution, identifying a bottleneck station in the process flow potential energy field, generating a task shunting scheduling instruction according to the potential energy gradient from the bottleneck station to the adjacent station, and executing the task shunting scheduling instruction.

Description

Intelligent clothing production ranking method and system based on Internet of things Technical Field The invention relates to the technical fields of industrial Internet of things and intelligent manufacturing, in particular to an intelligent ranking method and system for clothing production based on the Internet of things. Background As the garment manufacturing industry changes to flexible production modes of multiple varieties and small batches, the complexity of production line arrangement and dynamic scheduling becomes increasingly prominent. At present, the technical schemes in the field mainly comprise two main types, namely a semi-automatic scheduling system based on fixed rules or manual experience, wherein the system is highly dependent on personal experience of a scheduling person, complex procedures, personnel and equipment constraints are difficult to effectively quantize into a computable model, so that the subjectivity of a scheduling result is strong and reproducibility is poor, and an automatic scheduling system based on a traditional operation study optimization algorithm (such as linear programming and heuristic rules) is mainly classified into the system, and the system can realize automation to a certain extent, but has the core limitation that each scheduling is independent and an optimization process from zero is not effectively utilized, and the expected stability and efficiency of the scheduling scheme under similar production environments cannot be pre-estimated. The method has the main defects that firstly, a mechanism capable of automatically extracting, encoding and storing a reusable ranking mode (gene) from a historical successful case is lacked in a ranking scheme generation stage, so that each ranking is independently explored and effective experience which is verified cannot be inherited and reused, secondly, in a ranking scheme evaluation stage, only the current static resource matching degree is concerned, the adaptability of a scheme under different production environments (such as order emergency degree, personnel proficiency distribution and equipment health state) cannot be predicted and quantitatively evaluated by combining historical data, and thirdly, a scheduling mechanism capable of sensing the overall blocking situation of a production line in real time and driving dynamic reassignment of tasks according to situation gradients cannot be established in a scheme decision and execution stage, so that the system is enabled to respond to the sudden bottleneck in the production process with lag and adjust stiffness. Therefore, the invention provides an intelligent clothes production ranking method and system based on the Internet of things. Disclosure of Invention The invention provides an intelligent clothes production ranking method and system based on the Internet of things, which realize historical experience multiplexing by constructing a production mode genetic map library, perform multi-objective optimization by fusing cooperative adaptation degree and historical environment adaptation degree, and realize dynamic scheduling by a process flow potential energy field, so that ranking decisions are upgraded from static islands depending on experiences to data-driven intelligent closed-loop systems with historical learning and real-time self-balancing capabilities, and the defects are systematically solved. The invention provides an intelligent clothes production ranking method based on the Internet of things, which comprises the following steps: Step S1, retrieving at least one candidate genotype matched with the current order process characteristics and the production line resource state from a production mode genetic map library, wherein the production mode genetic map library stores genotype codes generated based on a historical successful ranking scheme, and the genotype codes at least comprise personnel skill feature vectors, equipment function feature vectors and logic relationship topology among working procedures; S2, analyzing the searched candidate genotypes into an initial ranking scheme, and calculating a collaborative fitness coefficient of a personnel-equipment-procedure combination distributed by each station in the initial ranking scheme, wherein the collaborative fitness coefficient is obtained by fusion based on a personnel skill quantization value, an equipment capacity quantization value and a procedure demand quantization value; step S3, constructing an optimized objective function which takes the overall collaborative fitness coefficient as a core, considers the historical environment fitness related to the candidate genotypes and meets the constraint of the intersection period and the load, and adopts a genetic algorithm to solve and output and execute an optimal ranking scheme based on the initial ranking scheme; And S4, dynamically constructing a process flow potential energy field according to the real-time task load and the processing ca