Operation Optimization in the Hot-Rolling Production Process

Abstract

The operation optimization problem (OOP) is an important issue in production process control and optimization in process industries, because the desired solution of OOP is the optimal setting for control variables, and this setting affects the product quality to a great extent. In this paper, the OOP in the hot-rolling production process of iron and steel industry is investigated. The OOP lies between the production scheduling layer and the process control layer in the integrated automation system in iron and steel industry, and its main task is to set the optimal values for control variables (i.e., rolling force, exit thickness, etc.) based on the production process constraints. These values are then set as the targets for the process control layer. Different from previous operation optimization of hot rolling process considered in the literature, the mathematical model constructed in this paper considers more practical process constraints such as the ramping constraints, to get high-quality products. To efficiently solve this problem, a hybrid self-adaptive genetic algorithm (HSaGA) is presented. The main feature of the HSaGA is that it utilizes four different crossover operators and can self-adaptively select the operators that are most appropriate for the current problem. Such a strategy can help to improve the convergence speed and robustness of the genetic algorithm. Besides this strategy, a solution selection strategy is also proposed to select parent solutions to perform the crossover operation. This strategy divides the entire population into two partsone with good quality in the objective function value and the other one with good quality in diversityand then the parent solutions are selected from them, to improve the search diversity during evolution and, thus, avoid being trapped in a local optimum. Experiments are carried out based on benchmark test problems and practical OOP of the hot-rolling production process. The computational results show that the proposed HSaGA is superior to many state-of-the-art evolutionary algorithms in the literature for benchmark problems and can obtain better results than the empirical method for the OOP of the hot-rolling process.

A Data-Driven Multiobjective Dynamic Robust Modeling and Operation Optimization for Continuous Annealing Production Process

“…In recent years, there have been many researches focused on the operation optimization problems in process industries such as the iron and steel industry and the chemical industry. [3][4][5][6][7][8][9][10][11] For the gas stirred ladle systems in the steelmaking process, Mazumdar et al 3) formulated a multiobjective optimization model with constraints to investigate inert gas injection in steelmaking ladles. In view of the hot stamping process, Mu et al 4) established the response surface model to optimize the heating parameters of hot stamping.…”

“…For the hot rolling production process, Jia et al 8) proposed the optimal design framework by using multiobjective optimization to obtain the optimal hot rolling parameters. Chen et al 9) established the operation optimization model of hot rolling process with the more practical constraints and presented a hybrid self-adaptive genetic algorithm to solve this model. Xia et al 10,11) developed a novel optimization method to solve the operation optimization problem of furnace temperature in the slab reheating process and proposed a mechanism and data analytics based hybrid model to study the furnace operation optimization problem.…”

“…In recent years, many kinds of multiobjective evolutionary algorithms (MOEAs) have been proposed in the literature. [19][20][21][22][23][24][25][26][27][28][29] Among these MOEAs, NSGA-II is one of the most widely adopted algorithms for practical optimization problems for process operation, 3,4,6,[8][9][10][11][27][28][29][30] due to its characteristics of fast computation speed and good convergence. Therefore, in this paper we prefer to adopt the NSGA-II as the solving method.…”

An Improved Particle Swarm Optimization for Selective Single Machine Scheduling with Sequence Dependent Setup Costs and Downstream Demands

“…In recent years, evolutionary algorithms have been widely adopted to solve both the continuous optimization problems ( [22][23][24]) and the combinatorial optimization problems derived in iron and steel industry ( [25][26][27]). The particle swarm optimization (PSO) algorithm is a kind of evolutionary algorithm proposed by Kennedy and Eberhart [28].…”

Control Vector Parameterization‐Based Adaptive Invasive Weed Optimization for Dynamic Processes

“…Furthermore, Zhou and Liu suggested a CVP method based on adaptive particle swarm optimization (APSO) by introducing a mutation into the APSO algorithm for chemical problems, where the convergence speed and accuracy of APSO are also greatly improved . Besides, many other algorithms have been widely adopted for dynamic optimization of chemical engineering problems .…”