Abstract: For the multi-contaminants batch water network optimization problem, a mixer-integer nonlinear programming model with the objective of minimum fresh water usage was established. Meanwhile, a solution strategy combined with matrix encoding genetic algorithm and trust region sequential quadratic programming(TRSQP)was proposed. In this paper, the matrix encoding method is easy to establish scheduling module, so the water reused relationship between the water units is more clearly defined.The integer variables was optimized by matrix encoding genetic algorithm and the continuous variables was optimized by TRSQP, the optimal strategy integrated the two methods to achieve the solution of MINLP model with the MATLAB software. The proposed method was used to solve two typical examples and the results obtained were superior to those in literature, which demonstrated that the approach introduced by this paper was effective.

Key words: water network, multi-contaminant, batch, genetic algorithm

摘要： 针对多杂质间歇用水网络优化设计问题,建立了以新鲜水用量最小为目标的混合整数非线性规划(MINLP)模型,并提出基于矩阵编码的遗传算法和信赖域序列二次规划(trust region sequential quadratic programming,TRSQP)法相结合的求解策略。本文采用矩阵编码方法为时间条件约束提供了方便,从而使用水单元之间水回用关系更加清晰明确;采用基于矩阵编码的遗传算法对整数变量进行优化,采用TRSQP法对连续变量进行优化,集成应用两种方法进行优化求解,从而获得MINLP模型的最优解;并借助MATLAB软件进行编程并实现。利用本文所提出的求解方法和策略对文献中2个典型案例进行求解,求解结果均优于文献数据。实例计算表明,本文所提求解方法是可行的。

关键词: 用水网络, 多杂质, 间歇, 遗传算法