Abstract: This paper proposes to use different scheduling methods and technologies to organically integrate to solve the problem of tobacco industry scheduling. Taking a cigarette manufacturing enterprise as the research and application background, the process flow and constraint rules in tobacco scheduling are analyzed, a double-layer architecture model of silk workshop scheduling is constructed, and the constraint rules in scheduling are embedded in the algorithm design process by using two-way scheduling scheduling technology and product optimization combination technology, and the algorithm design scheme is given, and the algorithm can be well adapted to the needs of enterprise silk thread scheduling through examples.
Keywords: tobacco processing; Flexible; Production scheduling; Smart scheduling
TS452.3 Document Identification Code A Article No. 0517-6611 (2017) 07-0081-02
Scheduling scheduling is the basis of the manufacturing system, and scheduling technology is of great significance to improve the production management level of enterprises, save costs, improve service quality, improve enterprise competitiveness and obtain higher economic returns. As a typical representative of the mixed process industry (the packing workshop is discrete, the silk workshop is the process type), the production characteristics of cigarette enterprises determine that their production process has various characteristics such as nonlinearity, randomness, and uncertainty, so their scheduling and scheduling problems are highly coupled mathematically. With the improvement of the automation level of cigarette enterprises, many enterprises have applied intelligent scheduling technology software to provide planning and guidance for actual production. The current scheduling technology software algorithms are mainly divided into two categories: first, the use of simulation means to simulate several scheduling strategies, and the optimal scheduling strategy is manually selected by production scheduling [1-2]; Second, based on the idea of modern graph theory, the process flow of the cigarette production line is abstractly modeled, so as to optimize the scheduling [3-4]. However, with the wide application of the characteristic process of "group processing" and the requirements of order-based production, the flexible processing line makes the production scheduling very complicated. The above algorithm cannot meet the needs of flexible scheduling goals and real-time scheduling scheduling in the workshop, resulting in the actual production scheduling plan still being completed manually and experienced. Limited manpower is difficult to ensure the accuracy of coordination and balance, which affects the productivity of cigarette production lines and is not conducive to the control of production costs, so it is urgent to conduct in-depth research on the scheduling and scheduling of flexible silk making threads in tobacco enterprises.
The author proposes a new idea of tobacco flexible yarn scheduling algorithm, that is, through the organic combination of two-way scheduling scheduling technology and product optimization combination technology, the constraint rules in scheduling are embedded in the algorithm design process, and the automatic scheduling of flexible yarn is realized through the application of hybrid algorithms, so as to realize the balanced processing and continuous production of the production line, and effectively control the production cost.
1 Overview of the silk making production process
1.1 Introduction to the silk processing process Taking a tobacco enterprise as an example, the silk making workshop is mainly composed of 2 blade processing lines, 3 leaf silk drying lines and 1 flavoring line. The equipment of the two blade treatment lines adopts a homogeneous configuration, and the tobacco leaf raw materials enter the leaf storage cabinet after loose moisture regain, feeding and moisturizing; The three leaf silk drying lines are respectively equipped with low, medium and high three kinds of different strength drying equipment, after drying, the leaf silk is mixed and mixed into the mixing cabinet, and the tobacco is added to the box-type storage silk inventory for storage. The processing process of the whole silk workshop has high flexibility, and the two blade processing lines can meet the parallel production of two blade group modules at the same time, and multi-module production needs can be realized through sequential processing; The leaf filament drying line can realize the parallel production mode of double specifications. The processing unit is flexibly connected, and the storage wire adopts the form of box storage, which improves the flexibility of silk making and coiling package production. The process flow is shown in Figure 1.
undefined
1.2 Scheduling rules and constraints The processing process of tobacco is different from the processing process of general mechanical products. In the production process of tobacco, due to its special process needs, there are many rules and constraints in the production process, some of which must be followed, some of which must be met as much as possible, and must be considered when carrying out production scheduling.
Tobacco storage capacity constraints: The total amount of tobacco stock of all brands cannot exceed the storage capacity of the entire box-type storage stock. Brand name processing priority constraints: When the production line switches from producing one specification of product to producing another specification of product, the switching cost depends on the order of switching, and the switching cost is lowest when switching from a high-specification product to a low-specification product. Equipment processing capacity constraints: In the simulation model of the production equipment of the process section, each production equipment has a rated processing capacity, and its daily processing capacity should meet the working time of the process section. Parallel production line balancing constraints: For parallel lines, their processing times should be synchronized to reduce energy consumption. Cache capacity constraint: The WIP cache capacity in each period of each process cannot exceed the inventory capacity limit. Cache time constraint: The WIP cache time cannot exceed the inventory time limit. Process path constraints: Specify processing equipment and paths for certain products. First-in-first-out (FIFO) principle.
2 Algorithm model design
2.1 Scheduling modeling The production and processing process of tobacco yarn has the characteristics of complexity and high degree of flexibility, and belongs to the category of gap batch processing process. Discrete decision-making variables and continuous decision-making variables exist simultaneously in the production process, and the system includes both continuous process variables, such as continuous changes in material flow; It also includes discrete process variables, such as the switching of production schemes, the introduction of random events, etc., so the production process is essentially a hybrid dynamic system. Due to the many constraints, the scheduling algorithm problem has become a very difficult non-deterministic problem to solve.
The flexible yarn scheduling algorithm adopts a hybrid optimization algorithm including mathematical programming, genetic algorithm, neural network algorithm and heuristic rule algorithm [5-8]. The algorithm model is designed as a two-layer architecture, and the upper layer is a scheduling optimization model with mathematical programming algorithms as the main body, nested into some neural network algorithms; The lower layer is a simulation model with the genetic algorithm as the main body, nested into some heuristic rules. The model architecture is shown in Figure 2.
The upper-level optimization model mainly makes decisions on the macroscopic production strategy in the scheduling cycle, such as the collection of production brand names, batch quantity, and brand name processing sequence. The goal of optimization model control is to reduce the cost of the production process under the condition of meeting the production conditions, raw material supply, product specification requirements and quantity demand constraints, and the upper model does not depict the processing details of batch products on the production line, such as the distribution of group modules in parallel production lines, materials in and out of storage cabinets, etc. Part of the neural network algorithm is embedded in the algorithm, and the product optimization combination production strategy is executed through the threshold activation function, so as to realize the continuous intensive production of the same brand. The lower simulation system carries out detailed product processing scheme planning through genetic algorithm optimization according to heuristic rules and various constraints, such as the distribution of each module in the production line and the sequential processing sequence. The product processing plan planning of the lower system will be fed back to the upper system, and the optimization model will be recalculated according to the new information and transmitted to the lower system, and the system will finally arrive at the processing execution plan after the end of the iteration.
2.2 Algorithm description The silk making line scheduling design is to meet the continuous production of the coil and baling workshop as the primary goal, and the silk batch operation demand plan is determined according to the machine work order of the coiling workshop and the inventory of finished tobacco. At the same time, it is necessary to pay attention to the execution sequence of batches, only reasonable order can ensure the continuity of reeling package production and ensure the improvement of the effective operation rate of the equipment. In the process of designing the flexible yarn scheduling algorithm, "pulling" and "pushing" are in parallel in two ways: pulling the roll operation plan according to the market order demand, the roll operation plan pulling to form the tobacco warehouse demand plan, and the specific silk production operation demand plan set is obtained through the dynamic inventory algorithm design and pulling; The silk making line operation plan adopts the "push" method for automatic scheduling, follows the principles of dynamic reconstruction of manufacturing resources and balanced equipment capacity, and optimizes and combines according to the production process sequence. The process steps of the automatic scheduling algorithm are as follows:
2.2.1 Environmental data information preparation. Read and store relevant environmental data information, including equipment, materials, time, strategy and other related scene data, and scheduling parameter setting information such as working days corresponding to the yarn and roll scheduling cycle. Read the initial state space of the day's silk thread, including the current capacity of the silk thread storage cabinet, mixing cabinet, tobacco library, process section-cabinet correspondence, brand-silk thread correspondence, etc.
2.2.2 Demand planning and scheduling of silk making operations. First, the schedule of the reel group is calculated to obtain the tobacco daily demand plan (Pi), defining P=(V,K,N), where V is the daily capacity of the coiling machine, K is the 10,000-piece yarn consumption rate, and N is the number of units. From the time of scheduling, after the T moment until the stock of tobacco is consumed, there are:
undefined
In the formula, Q0 is the amount of tobacco stock, so that the inventory holding time T is sorted, and the shorter the inventory maintenance time, the higher the priority.
Secondly, optimize the product brand name combination. Define the product optimization combination function C(t)=ψ[θ(t)], determine the current actual tobacco inventory Qk of each brand name at t moment, and when the current actual tobacco inventory Qi of the brand name reaches the threshold θi, the production brand name is combined and optimized.
Finally, identify the collection of job requirement plans. Set the regular stock Qi of the tobacco library of each brand name, and soften the rigid limit of the stock of brand names by relaxing variables (Qiu, Qid). Define Hi(t)=[Qiu(t)Qid(t)], and the system obtains the set of silk yarn operation demand plans R, R=(p,n) based on Hi(t) and the current actual tobacco inventory Qk, where p is the production brand name and n is the batch quantity.
2.2.3 Schedule of silk making operations. The system adopts the push method to automatically schedule production in the order of photo cigarette treatment section - leaf silk drying section - flavoring section. Firstly, the set R of the demand plan of the filament making operation is read, and the rules such as brand name switching cost, resource grouping, balanced production, and working hour constraint are used to schedule production, so as to obtain the set of production planning of the tobacco treatment section ∑Y1(p,n). Secondly, the scheduling plan of the sheet tobacco treatment section was read ∑Y1 (p,n), and the principles of process path constraint, parallel production line efficiency constraint, working hour constraint, FIFO and other principles were called to schedule production, so as to obtain the scheduling plan set of leaf drying section ∑Y2( p, n). The third is to read the leaf silk drying line scheduling plan Y2(∑p,n), and schedule the production according to the principles of working time constraint and FIFO, so as to obtain the ∑ Y3(p,n) of the scheduling plan of the perfumed section. Finally, according to the principle of minimizing production costs, optimization is carried out, and if the target is not met, iterative rearrangement is carried out, and finally the daily operation plan of yarn yarn Y1(p,n), Y2(p,n), Y3(p,n) is obtained.
3 Application examples
Taking the roll production plan of a cigarette factory on a certain day as an example, the flexible silk making line consists of 2 parallel tobacco sheet processing lines, 3 parallel leaf drying lines (2 sheet drying lines, 1 airflow drying line) and 1 tobacco flavoring line, of which the brand names B, C and D are all grouped products. Its scheduled operation process is as follows: brand A plans to produce 360 boxes; Brand B plans to produce 720 boxes; Brand C plans to produce 540 boxes; Brand D plans to produce 450 cases.
The system first calculates the daily tobacco demand for each brand name; Secondly, the current actual tobacco inventory of each brand name was judged, and it was found that the current actual tobacco inventory of D brand name reached the threshold θd, and the production brand names on this day were A, B and C; Finally, combined with the dynamic tobacco inventory algorithm, the set R, R=[(A,2-3) (B,5-6) (C,3-4)] of the yarn operation demand plan is obtained. After the demand plan is determined, the program automatically runs to obtain the scheduling plan set of each section, and calls the objective function F for optimization according to the principle of minimizing production costs, so as to obtain the specific daily operation plan of the silk making thread: 1# piece of tobacco treatment line is A1Ba6Cb3, 2# piece of tobacco treatment line is A1Bb6Ca3, 1# sheet drying line is A2Ba1Ca1, 2# thin plate drying line is Ba5Ca2, 3# air drying line is Bb6Cb3, tobacco flavoring line is A1B2A1B4C3.
The actual operation results show that all batches are completed within the demand time, which not only meets the production needs of coil and package, but also reduces the number of brand changes, the production line runs efficiently and consumably, and the scheduling algorithm has good feasibility.
4 Conclusion
The automatic scheduling of flexible silk yarn in tobacco enterprises is a difficult point. This study uses hybrid algorithms to carry out the research and practice of scheduling and scheduling, solves the shortcomings of some single technical methods, realizes the intelligence of scheduling, the efficiency of scheduling results and the saving of production costs, and provides new ideas for in-depth research on the scheduling.
bibliography
[1] GUO Jintian. Optimization and simulation of production scheduling of automobile mixed flow assembly line[D].Wuhan:Wuhan University of Technology,2008.)
[2] SUN Jiexiang. Research and implementation of silk production scheduling method for tobacco industry based on simulation[D].Beijing:General Academy of Mechanical Sciences,2009.)
[3] CHEN Zhigang, SUN Shuqing. Design of automatic scheduling system for silk making wire in cigarette factory[J].Mechanical Engineer,2009(11):105-107.)
[4] LI Tao, GUAN Zhong, MA Lu. Tobacco Science and Technology,2006912):20-23.)
[5] Jin Jian, Jin Zhao, Qi Yuedong. Establishment and optimization of cigarette production planning and scheduling model[J].Computer Engineering and Applications,2013,49(18):253-259.)
[6] HU J.MES in the cigarette industry in the productionprocess and quality control of application and analysis [D]. Nanjing University of Science and Technology , 2004.
[7] ZORYK-SCHALLA A J,FRANSOO J C,KOK T D,et al. Modeling the planning process in advanced planning systems [J]. Information & management , 2004 , 42 (2) :75-87.
[8] PAM F C,SHI H B,SONG H.Scheduling Algorithm Research on Processing Line of Cigaarette Plant[C]//International Conference on Future Computers in Education (ICFCE 2011). Sham Shui Po: Int Industrial Electronic Center,2011.
Transferred from Wanfang Data, Xu Yonghu, if there is infringement contact deleted.
Follow the subscription account "Smart Manufacturing Trends" for more industry information.