论文部分内容阅读
汽油调合调度是炼油工业中的一个重要问题。一方面,该问题具有调度问题本身所具有的组合优化特性;另一方面,由于调合各种具有不同品质特性的物料,导致该优化问题的非凸性。本文提出一种新型的全局优化算法,用于求解基于连续时间汽油调合调度模型的混合整数非线性规划问题。该模型包含调合配方优化、分配问题及若干操作特性和约束;算法上采用分段Mc Cormick松弛(PMCR)和规范多参数解聚(NMDT),计算全局最优解的估计值,其松弛技术将双线性项中的一个变量值域进行分割,进而在每一个分段上产生凸松弛;通过增加分段数和缩减变量的值域,提高对全局最优解的估计。本文利用该算法求解四个案例,并与两个商业全局优化求解器和两个启发式算法进行比较,结果表明,本文提出的全局优化算法与商业求解器具有同等水平,但是在计算速度上稍逊于启发式算法。
Gasoline blending scheduling is an important issue in the refining industry. On the one hand, the problem possesses the combination optimization characteristic of the scheduling problem itself; on the other hand, the non-convexity of the optimization problem results from the blending of various materials with different quality characteristics. In this paper, a new global optimization algorithm is proposed to solve the mixed integer nonlinear programming problem based on continuous-time gasoline blending scheduling model. The model includes the optimization of blending formula, the distribution problem, and some operating characteristics and constraints. The algorithm uses piecewise Mc Cormick relaxation (PMCR) and canonical multi-parameter depolymerization (NMDT) to calculate the global optimal solution. The relaxation technique In this paper, a value range of a variable in bilinear terms is divided, and then convex slack is generated in each segment. The estimation of the global optimal solution is improved by increasing the number of segments and reducing the range of variables. In this paper, the algorithm is used to solve four cases and compared with two commercial global optimization solvers and two heuristic algorithms. The results show that the global optimization algorithm proposed in this paper has the same level with commercial solvers, but the computational speed is slightly Less than the heuristic algorithm.