Abstract: Intelligent optimization methods are playing an increasing role in dynamic optimization in chemical engineering due to their simplicity and good global search ability. Unfortunately, traditional intelligent optimization methods often suffer from a relatively slow convergence rate. An iteratively adaptive particle swarm optimization (IAPSO) approach for dealing with general chemical dynamic optimization problems was proposed. A dynamic optimization problem was first converted into a nonlinear programming (NLP) problem through control parameterization (CP); then the proposed IAPSO approach was used to solve the NLP. The proposed IAPSO approach had a faster convergence rate than the conventional particle swarm optimization approach mainly for two reasons: IAPSO updated parameters adaptively according to population distribution characteristics; IAPSO executed adaptive particle swarm optimization algorithm iteratively by reducing search space to get a more accurate solution. Several benchmark dynamic optimization problems were explored as illustration and the results showed that the proposed IAPSO approach was simple, efficient, and considerably outperformed the conventional PSO method in terms of convergence rate.

Key words: dynamic, optimization, iteratively adaptive particle swarm optimization, region reduction, reactor

摘要： 智能优化方法因其简单、易实现且具有良好的全局搜索能力，在动态优化中的应用越来越广泛，但传统的智能方法收敛速度相对较慢。提出了一种迭代自适应粒子群优化方法（IAPSO）来求解一般的化工动态优化问题。首先通过控制变量参数化将原动态优化问题转化为非线性规划问题，再利用所提出的迭代自适应粒子群优化方法进行求解。相比传统的粒子群优化方法，该种迭代自适应粒子群优化方法具有收敛速度更快的优点，主要原因是：该算法根据粒子种群分布特性自适应调整参数；该算法通过缩减搜索空间并迭代使用粒子群算法搜索最优解。将提出的迭代自适应粒子群方法应用到多个经典动态优化问题中，测试结果表明，该方法简单、有效，精度高，且收敛速度比传统粒子群算法有显著提升。

关键词: 动态, 优化, 迭代自适应粒子群, 区域缩减, 反应器