CN-122022388-A - Novel intelligent high-quality steel wire scheduling method based on big data model

CN122022388ACN 122022388 ACN122022388 ACN 122022388ACN-122022388-A

Abstract

The invention relates to the technical field of big data and production plans, and discloses a novel intelligent high-quality steel wire scheduling method based on a big data model, which comprises the steps of extracting material characteristic parameters of each steel order to construct a material process knowledge graph; the method comprises the steps of extracting material volatility indexes and order stability indexes of all customer orders, marking the customers, calculating process window coincidence degrees from order to order, taking the process window coincidence degrees as constraints, dynamically calculating residual available capacity of all production lines, taking the process window coincidence degrees and the residual available capacity as constraints, determining and grading combinations of the orders and the production lines meeting production constraints, generating a production scheme aiming at the orders to be produced according to grading results, executing production according to the production scheme, and carrying out self-adaptive adjustment on pull speed of the current production line. The invention can realize accurate configuration and dynamic optimization of production line resources aiming at the characteristics of orders of multiple varieties and small batches in the manufacturing industry of special steel wires.

Inventors

BIAN JIANZHONG
LIU YANG
CAO LEI
JING DAN
CHEN WEI
LI HAIYAN
YANG FAN
CHEN XIAOLIANG

Assignees

江苏永钢集团有限公司
联峰钢铁（张家港）有限公司

Dates

Publication Date: 20260512
Application Date: 20260409

Claims (10)

1. The novel intelligent high-quality steel wire scheduling method based on the big data model is characterized by comprising the following steps of: Extracting material characteristic parameters of each steel order from a production database, and constructing a material process knowledge graph based on the material characteristic parameters, wherein the material process knowledge graph comprises a production line digital twin body and an order feature vector; based on the material process knowledge graph, extracting a material volatility index and an order stability index of each customer order, generating a cross classification result through two-dimensional classification to mark customers, and outputting a demand prediction result; taking the predicted orders in the demand prediction results as to-be-scheduled production orders, calculating process window coincidence degrees from order to order production line to line, and dynamically calculating the residual available capacity of each production line by taking the process window coincidence degrees as constraints; determining an order and production line combination meeting production constraint by taking the process window conformity and the residual available capacity as constraint conditions, comprehensively scoring the order and production line combination based on a multi-dimensional evaluation index, and generating a scheduling scheme for the order to be scheduled according to a scoring result; And when any index is lower than a corresponding threshold value, carrying out self-adaptive adjustment on the pull speed of the current production line, and feeding the production data back to the production line digital twin body.
2. The new intelligent high-quality steel wire production method based on the big data model as claimed in claim 1, wherein the construction material process knowledge graph comprises the following steps: Extracting carbon content, alloy element proportion, phase transition temperature and surface quality requirement grade of each steel order from a production database as material characteristic parameters, and converting each order into a multidimensional order feature vector; The method comprises the steps of synchronizing equipment states, process capacities and historical production data of all production lines in real time, and establishing a production line digital twin for each production line, wherein the production line digital twin comprises equipment precision, an energy consumption curve, a historical yield and a material compatibility matrix used for representing preparation time required for switching from any steel grade to another steel grade; And calculating the material similarity of the order feature vectors of any two orders in each dimension based on the order feature vectors and the production line digital twin body, and marking the order combination with the material similarity larger than a preset threshold as a high-compatibility order pair so as to establish an index mechanism based on the material similarity and generate a material process knowledge graph.
3. The new intelligent special steel wire scheduling method based on the big data model according to claim 2, wherein the converting each order into a multidimensional order feature vector comprises: Extracting specific numerical values of the carbon content, the alloy element proportion, the phase transition temperature and the surface quality requirement level of each order according to a preset dimension sequence by taking the material characteristic parameters as vector dimensions, and combining to generate a one-dimensional parameter sequence as the multidimensional order feature vector; The establishment of the line digital twin for each line comprises: The method comprises the steps of constructing a static process capacity frame comprising equipment precision, the energy consumption curve, the historical yield and the material compatibility matrix, collecting equipment state parameters and production data of roller abrasion states of all production lines and heating furnace working conditions in real time, and synchronously updating the equipment state parameters and the production data into the static process capacity frame to generate the production line digital twin body.
4. The new intelligent high-quality steel wire production method based on the big data model as claimed in claim 2, wherein the material compatibility matrix is obtained by the following steps: constructing a two-dimensional initial matrix structure according to all steel types contained in a production history database, wherein the rows of the initial matrix structure correspond to the initial steel types before switching, and the columns correspond to the target steel types after switching; Extracting preparation time required by actual switching between different initial steel grades and target steel grades, wherein the preparation time is accumulated time consumption of roller adjustment, heating furnace working condition switching and head piece inspection; and carrying out statistical calculation on the preparation time under the switching path from the same initial steel grade to the target steel grade, taking average time consumption as basic preparation time after removing abnormal values, and filling the basic preparation time corresponding to each steel grade combination as matrix elements into the corresponding row-column intersection positions of the initial matrix structure to obtain the material compatibility matrix.
5. The new intelligent high-quality steel wire scheduling method based on the big data model according to claim 1, wherein the generating the cross classification result by the bimodal clustering to mark the clients comprises: Calculating the deviation of the historical orders and the historical average of the clients in each parameter dimension based on the material characteristic parameters of each client historical order to obtain a material volatility index, and dividing the clients into low-difference classes, medium-difference classes and high-difference classes according to a preset first threshold interval to obtain a first clustering result; Based on historical transaction records of all clients, calculating comprehensive stability degree of delivery frequency and order quantity to obtain order stability indexes, and dividing clients into stable classes, general classes and fluctuation classes according to a preset second threshold interval to obtain a second aggregation result; and carrying out cross combination on the first clustering result and the second clustering result to obtain the cross classification result, marking the clients simultaneously belonging to the low-difference stable class as low-difference stable clients, and marking the clients simultaneously belonging to the high-difference stable class as high-switching-cost clients.
6. The new intelligent special steel wire scheduling method based on the big data model as claimed in claim 1, wherein the order-by-order production line-by-production line computing process window conformity comprises: based on the production line digital twin bodies, extracting equipment precision, a roller abrasion state and a heating furnace working condition as components, and constructing a process capability vector of each production line; extracting the carbon content, the phase transition temperature and the surface quality requirement level in the material characteristic parameters of the order to be produced, and establishing a mapping relation between the material characteristic parameters and the process capability vector dimension by dimension; Calculating a compatibility function value under each dimension based on the mapping relation, and performing continuous multiplication processing to obtain a process window conformity; the dynamically calculating the remaining available capacity of each production line comprises: combining and removing orders and production lines with the process window coincidence degree lower than a preset coincidence degree threshold; marking the order which is confirmed to be unchanged as a locked order, and extracting the capacity occupied by the locked order on each production line; after deducting the capacity occupied by the locked order, calculating the residual available capacity of each production line for the to-be-scheduled order in a planning period according to the combination of the non-removed order and the production line.
7. The new intelligent special steel wire scheduling method based on the big data model according to claim 6, wherein the generating process of the scheduling scheme of the order to be scheduled comprises the following steps: Determining that the process window conformity reaches a preset conformity threshold and that the order and line combination with the residual available capacity larger than a preset capacity value is a feasible combination; For each feasible combination, extracting evaluation indexes of quality loss, processing cost, ton steel energy consumption and switching time, and determining objective weights of the evaluation indexes by adopting an objective weighting method; Based on objective weights of the evaluation indexes, performing comprehensive score calculation on all feasible combinations with the residual available capacity larger than a preset capacity value to obtain a comprehensive score matrix, and sequencing the comprehensive score matrix from high score to low score; Selecting the feasible combinations with scores higher than a preset proportion in the comprehensive scoring matrix as seed individuals of an initial population of a multi-objective optimization algorithm, and randomly generating other individuals to generate a scheduling scheme; The scheduling scheme comprises a schedule of the locked orders, candidate scheduling sequences of non-locked orders, and basic pull rates and corner safety threshold intervals corresponding to the candidate scheduling sequences.
8. The new intelligent high-quality steel wire scheduling method based on the big data model according to claim 7, wherein the non-locking order comprises a candidate scheduling order with a dynamic adjustment margin, the dynamic adjustment margin comprises a movable time window and a replaceable production line; The step of performing comprehensive scoring calculation to obtain a comprehensive scoring matrix comprises the following steps: based on the objective weight of each evaluation index, calculating the relative proximity between each feasible combination and an ideal optimal solution; And taking the relative proximity as a comprehensive score, and constructing the comprehensive score matrix aiming at the to-be-scheduled order.
9. The new intelligent special steel wire production method based on the big data model according to claim 7, wherein the real-time acquisition of production data and calculation of material stability index and equipment health index comprises the following steps: Acquiring molten steel component deviation, superheat degree and crystallizer liquid level fluctuation data of a current production line in real time, and obtaining a material stability index for representing the stability degree of input conditions through comprehensive calculation; Collecting the vibration amplitude of the foot roller, the blockage rate of the nozzle in the secondary cooling area and the taper deviation data of the crystallizer of the current production line in real time, and obtaining the equipment health index for representing the cooling uniformity through comprehensive calculation; And inverting the distribution of the temperature field at the corner of the casting blank produced by the current production line in real time based on the infrared temperature measurement array, calculating the temperature uniformity index at the corner and the surface crack detection rate, and generating a quality index for representing the consistency of the production result.
10. The new intelligent special steel wire scheduling method based on the big data model according to claim 9, wherein the adaptive adjustment of the pull speed of the current production line comprises: when the material stability index or the equipment health index is lower than a preset first safety threshold, acquiring a basic pulling speed and a corner safety threshold interval set by the production scheduling scheme; The method comprises the steps of acquiring historical production data, calibrating material influence weights and equipment influence weights by carrying out regression analysis on the historical production data, matching the material stability index with the material influence weights, and matching the equipment health index with the equipment influence weights; Based on the material influence weight, the equipment influence weight and a nonlinear compensation factor corresponding to the current corner state, carrying out dynamic compensation calculation on the basic pull speed, and generating an adjusted actual pull speed; The nonlinear compensation factor is used for controlling the active deceleration of the current production line to inhibit the formation of corner cracks when the current corner state approaches the lower limit of the corner safety threshold interval.

Description

Novel intelligent high-quality steel wire scheduling method based on big data model Technical Field The invention relates to the technical field of big data and production planning, in particular to a novel intelligent high-quality steel wire scheduling method based on a big data model. Background Along with the rapid development of technology and the continuous change of market demands, the automation degree of a production line is continuously improved, and the traditional production scheduling method cannot meet the demands of the modern manufacturing industry. The traditional scheduling method mainly depends on historical data and static prediction, and cannot accurately reflect dynamic changes of markets and fluctuation of customer demands, so that production efficiency is low and stock backlog is caused. In addition, most of the existing scheduling systems adopt manual statistics and fixed scheduling strategies, so that the customer order data are difficult to fully utilize and analyze, and the defects particularly exist in the aspect of demand prediction for distinguishing stable customers from fluctuating customers. In the manufacturing industry of special steel wires, the types of orders are characterized by multiple varieties and small batches, the types of products are various, and the requirements of each order can be different. This complex production environment places higher demands on the flexibility and adaptability of the production line. However, most of the existing production lines and products are tightly bound, so that the production lines are difficult to flexibly configure, the problems of low equipment utilization rate, long production period and incapability of timely meeting customer exchange period requirements are easily caused. In order to solve these problems, the industry has begun to explore intelligent optimization methods based on big data models to improve production efficiency and meet different user demands. However, how to accurately process and optimize the production of the customer order data, so as to realize the efficient operation of the production line and the maximum utilization of resources is still a technical problem to be solved currently. Particularly in distinguishing customer types, analyzing order characteristics, and adaptively adjusting production strategies based on such information, there is still significant room for improvement. For the problems in the related art, no effective solution has been proposed at present. Disclosure of Invention Aiming at the problems in the related art, the invention provides a new intelligent high-quality steel wire scheduling method based on a big data model, so as to overcome the technical problems in the prior art. For this purpose, the invention adopts the following specific technical scheme: A new intelligent high-quality special steel wire production method based on a big data model comprises the steps of extracting material characteristic parameters of each steel order from a production database, constructing a material process knowledge graph based on the material characteristic parameters, wherein the material process knowledge graph comprises a production line digital twin body and order feature vectors, extracting material volatility indexes and order stability indexes of each customer order based on the material process knowledge graph, generating a cross classification result through two-dimensional classification to mark customers, outputting a demand prediction result, taking the prediction order in the demand prediction result as an order to be produced, calculating process window coincidence degree according to the order line, taking the process window coincidence degree as constraint, dynamically calculating residual available capacity of each production line, taking the process window coincidence degree and the residual available capacity as constraint conditions, determining an order and production line combination meeting production constraint, comprehensively scoring the order and the production line combination based on a multi-dimensional evaluation index, generating a production scheme aiming at the order to be produced according to the scoring result, executing production according to the production scheme, real-time producing the material stability indexes and equipment, and feeding back the material stability indexes to the twin line health index to the current digital data when the corresponding to the twin line is lower than a threshold value, and self-adapting to the current production index. The beneficial effects of the invention are as follows: (1) The invention realizes the dynamic monitoring and real-time analysis of the customer demands by utilizing the customer order data in the production database and combining with the big data analysis technology, overcomes the limitation that the traditional method only depends on historical data and static prediction, and can more accurately