论文部分内容阅读
遗传算法具有很好的全局寻优能力,但在接近最优点时易波动,使得寻优时间加长;可变容差法作为很好的局部寻优算法,却对初始点要求较高。针对这两种算法的优缺点,设计出以遗传算法作为初始优化方法,在满足一定收敛条件后,使用可变容差法进一步寻优的两步式优化方法。经过Rosenbrock函数优化验证,证明这种两步式优化算法克服了原有单一算法的缺点。以流程模拟为基础,仿真验证了常压蒸馏塔能耗目标的操作优化,达到了预期效果,具有较好的实践意义。
Genetic algorithm has good global optimization ability, but it is easy to fluctuate near the optimal point, which makes the optimization time longer. As a good local optimization algorithm, the variable tolerance method requires a higher initial point. Aiming at the advantages and disadvantages of these two algorithms, a two-step optimization method using genetic algorithm as the initial optimization method and satisfying the certain convergence condition and further optimizing using the variable tolerance method is designed. After Rosenbrock function optimization verification, this two-step optimization algorithm is proved to overcome the shortcomings of the original single algorithm. Based on the flow simulation, the operation optimization of energy consumption target of atmospheric distillation column was verified by simulation, which achieved the expected results and had good practical significance.