Multi-objective Optimization Algorithms And Their Applications In Chemical Engineering

Multi-objective optimization algorithm is widely used in the chemical industry, such as process optimization and operational control, chemical equipment design, environmental engineering, etc. Therefore, looking for ways to solve the multi-objective optimization problem(MOP) have significant value in scientific and engineering fields. In recent years, more and more scholars optimize cases of chemical process by combining multi-objective optimization algorithms with process simulator. Due to complexity of the process simulation, that requiring a lot of time for convergence, optimization algorithms must quickly achieve convergence in the case of less objective function evaluations. NSGA-II(Non-dominated Sorting Genetic Algorithm II) is the most widely used algorithm to solve MOP, but it requires ten thousands or more evaluations of the objective functions to get better results. And NSGA-II has some other shortcomings, such as ‘premature' and weak local search ability. Thus, this paper aims to propose an efficient multi-objective optimization algorithm and apply it to the optimization of chemical processes. The main work in this paper are as following:(1) Described the multi-objective optimization algorithm research background and significance, and introduced the development of multi-objective evolutionary algorithm in research and engineering applications. Introduced the research and application development of LCA(Line-up Competition Algorithm);(2) Introduced some concepts and definitions of MOP, then, presented the calculation process and key operators of LCA and NSGA-II. Studied measures of algorithm performance, included the metric ? to measure the extent of convergence and the metric?to measure the extent of spread solutions;(3) Proposed a multi-objective optimization algorithm named MOLCA(Multi-objective Line-up Competition Algorithm), using a variety of strategies to reduce the time of evaluations of the objective function at the same time to achieve convergence fast. The setting of MOLCA main parameters was discussed, and then used the classic test function to exam and analyze MOLCA. Compared with NSGA-II, MOLCA performed better. Applied MOLCA to the optimization of FCC main fractionation column operation parameter, the Pareto sets were given to choose;(4) To solve the drawbacks of premature convergence and low computational efficiency of NSGA-II, a hybrid algorithm was proposed which couples MOLCA and NSGA-II, namely MOLCA-NSGA-II. It performed better in convergence, distribution and operating efficiency on the test problems than NSGA-II. Double objective of recovery of olefins and energy consumption in MTO separation process are optimized by MOLCA-NSGA-II on the Aspen Plus and MATLAB integration platform, which providing meaningful guidance for the practical operation.