| Hot rolling production process in modern large steel plants is constituted by hot rolling machine and multiple parallel walking reheating furnaces. An inventory slabs or a high-temperature slab directly casted by a two-strand caster is firstly heated by a selected walking reheating furnace. The slab is discharged from the walking reheating furnace when its temperature reaches a specified temperature, then it is delivered to hot rolling machine, and converted into a certain length, width, thickness, hardness, and surface quality of coil by roughing rolling, finishing rolling and coiling. A group of slabs rolled by the same set of finishing rolling work rollers is called a rolling unit.Hot rolling production planning (HRPP) includes rolling planning (RP) problem and reheating furnaces charging planning (RFCP) problem, RP includes slab selecting, DHCR single mode rolling unit planning (DSUP), NO-DHCR single mode rolling unit planning (NDSUP) and NO-DHCR mix mode rolling unit planning (NDMUP) problems. HRPP is firstly to select some slabs for making rolling units from a lot of given slabs:inventory slabs and waiting produced slabs, then make DHCR single mode, NO-DHCR single mode and NO-DHCR mix mode rolling units, and determine the heating facility, the start time and the end time of each slab in rolling units for constructing the production timetable of walking reheating furnaces (i.e., reheating furnaces charging plan) with hot rolling plan as basis.The quality of slab selecting affects delivery data, the quality of rolling unit planning affects production cost, material supplies for downstream production lines, product quality and productivity, but the quality of RFCP affects energy consumption. The constraints about the surface quality requirement and the quantity of staple material slabs rolled consecutively with the same steel type have many kinds of possibility description, and they can be expressed as multiple constraint equations. So the DSUP, NDSUP and NDMUP problem cannot be solved using existing optimization method. The constraint about the quantity of slabs processed simultaneously on each walking reheating furnace also has many kinds of possibility description, so the RFCP problem cannot be solved using existing optimization method. The disadvantages of the existing research literatures are as follows:(1)the existing research methods for slab selecting is difficult to be applied in practice because of neglecting downstream production requirements, (2)the existing research methods for DSUP is difficult to be applied in practice because the width of each waiting produced slab is pre-determined, and the integrated production with two strand caster and hot rolling machine is neglected, (3)the existing research methods for NDSUP and NDMUP is difficult to be applied in practice because the region of each slab in rolling unit is pre-determined, and the surface quality requirement of coils is neglected, (4)the existing research methods for RFCP is difficult to be applied in practice because of neglecting the capacity constraint of each reheating furnace. At present the steel plant has to use inefficient manual method. For hot rolling plan obtained b manual method, the average length of rolling units is smaller, the weight of each flow slabs i units is often far away from its goal weight and even out of its weight range. So it increase cost, reduces production efficiency, and affects downstream regular production because c raw material supply problem. Walking reheating furnaces charging plan obtained by manua method increases energy consumption.This dissertation takes the hot rolling production line of a large steel plant with three walkin reheating furnaces and one hot rolling machine as the research background. Under the researc background, the methods for rolling planning and reheating furnaces charging planning an proposed. The major contributions are listed as follows.1.The slab selecting method:On the basis of analyzing the considered factors includin weight and delivery date etc, the performance indexes and the constrains of the problem an described. Considering the difficulty in accurately formulating the slab selecting problem, heuristic method for the slab selecting problem is proposed by combining expert knowledge.2.The DSUP method:A model for the DSUP problem is established. The objectiv functions are to minimize the difference between two strand time for casting waiting produce slabs of each rolling unit, to minimize the total jumps between adjacent slabs of rolling unit; maximize total priorities of slabs of units, maximize the average rolling length of units, an minimize the difference between the rolling and casting sequences of waiting produced slab of each rolling unit. The decision variables are the slabs of each rolling unit, the region c each slab in rolling unit, the rolling sequence of slabs of each rolling unit, the width of eac waiting produced slab of each rolling unit, the casting strand of each waiting produced slab c each rolling unit and the casting sequence of waiting produced slabs of each rolling unit. Th constraint equations are that each rolling unit only contains a steel grade of waiting produce slabs, the weight of waiting produced slabs of each rolling unit is more than the specifie weight given by the steel grade of waiting produced slabs, and the total length of a staple sla and its former staple slabs is not more than the specified length given by the steel typle of thi slab in the same rolling unit. Since the problem cannot be solved by existing optimizatio methods, a method based on a hybrid of variable neighborhood search and ant colon optimization is proposed. The method includes two algorithms:an algorithm based o heuristics for war-up planning and an algorithm based on a hybrid of variable neighborhoo search and ant colony optimization for staple planning. In iterative processes of the hybri algorithm, the staple material of each rolling unit is roughly obtained, then the casting strano the casting sequence and the width of each waiting produced slab constituting staple materia in each unit are obtained, finally the rolling sequence of each waiting produced slab in eac unit is adjusted. The experiment with industrial data shows that the result obtained by th proposed hybrid algorithm is superior to the result obtained by manual method.3.The NDSUP method:A model for the NDSUP problem is established. The objectiv functions are to minimize the total jumps between adjacent slabs of rolling units, maximiz total priorities of slabs of units, and maximize the average rolling length of units. The decisio variables are the slabs of each rolling unit, the region of each slab in rolling unit and the rolling sequence of slabs of each rolling unit. The constraint equations are that each rolling unit only contains a steel type of staple slabs, and the total length of a staple slab and its former staple slabs is not more than the specified length given by the steel typle of this slab in the same rolling unit. Since the problem cannot be solved by existing optimization methods, a method based on a hybrid of simulated annealing and ant colony optimization is proposed. The method includes two algorithms:an algorithm based on heuristics for war-up planning and an algorithm based on a hybrid of simulated annealing and ant colony optimization for staple planning. The staple planning experiment with industrial slab data shows that the proposed hybrid algorithm is superior to single ant colony optimization algorithm and single simulated annealing algorithm in terms of precision, but its solving time is much less than manual time.4.The NDMUP method:A model for the NDMUP problem is established. The objective functions are to minimize the difference between the weight of each flow slabs in units and its target weight, minimize the total jumps between adjacent slabs of rolling units, maximize total priorities of slabs of units, and maximize the average rolling length of units. The decision variables are the slabs of each rolling unit, the region of each slab in rolling unit and the rolling sequence of slabs of each rolling unit. The constraint equations are that the steel types of adjacent staple slabs of each rolling unit are compatible, the quantity of staple slabs rolled consecutively with the same steel type is not less than the specified quantity given by the steel type in the same rolling unit, and the total length of a staple slab and its former staple slabs is not more than the specified length given by the steel type of this slab in the same rolling unit. Since the problem cannot be solved by existing optimization methods, a variable neighborhood tabu search algorithm is proposed. Search strategy for improving searching efficiency is given based on the neighborhood structures:deletion, insertion, exchange, swap and section swap in the algorithm. The experiment with industrial slab data shows that the result obtained by the proposed algorithm is superior to the result obtained by manual method.5.The RFCP method:A mathematical model for the RFCP problem is established. The objective functions are to minimize the practical heating time of slabs, minimize the waiting time for heated slabs during hot rolling, and minimize the number of the mix charges of cold and hot slabs. The decision variables are the heating facility, the start time and the end time of each slab. The constraint equations are that the processing time of each slab on the furnace is more than the specified minimum processing time, slabs are one by one discharged from walking reheating furnaces in the rolling sequence, and the quantity of slabs processed simultaneously on each furnace does not exceed the specified quantity. Since the problem cannot be solved by existing optimization methods, an algorithm based on ant colony optimization is proposed. The heating facility of each slab are directly optimized by the algorithm, but the start time and end time of each slab are obtained by a heuristic method based on constraints. The experiment with industrial slab data shows that the result obtained by the proposed algorithm is superior to the result obtained by manual method.6.The system computational experiment for the above proposed five methods is done using a group of industrial slab data. The rolling plan obtained by the proposed methods is compared with the rolling plan obtained by manual method, the average length of rolling units rises from 55.376kn to 68.401km, the average weight of rolling units rises from 1184.9t to 1460t, the average of the deviation between the weight of arranged slabs corresponding each flow and its target weight drop; from 135.1t to 115.2t. The proposed method obtains the rolling plan within 244 seconds, whil(?) manual method needs about 3 hours. The walking reheating furnaces charging plan obtained by the proposed method is compared with the walking reheating furnaces charging plan obtained by manual method, the average heating time of slabs saves 3 minutes. |
PDF Full Text Request