Abstract:

The larger the scale of heat exchange network, the extreme points in its solution space increase exponentially. The optimization requires not only the algorithm to have strong global optimization ability, but also the high-precision search of local solution space is indispensable. Since random walk algorithm with compulsive evolution (RWCE) is difficult to take into account the local search ability when optimizing large heat exchange networks, it is easy to miss the optimal solution. In order to increase the population size of the algorithm, a parallel double-layer RWCE algorithm was proposed by combining fine search and parallel computing. Based on multi-core parallel technology, the algorithm establishes the basic-search layer and fine-search layer through parallel thread allocation. With the support of parallel computing technology, the global search ability of the algorithm was greatly improved in the basic-search layer. The fine-search layer carried out real-time fine search for the current optimal solution from the basic-search layer, avoiding the phenomenon that the imperfect solution replaced the optimal solution. Finally, through two examples, the results showed that the parallel double-layer RWCE algorithm not only has stronger global search ability, but also has high-precision local search ability, and effectively protects the optimal solution in the optimization process.

Key words: process systems, heat exchange network, optimization, algorithm, double-layer algorithm, parallel computing

摘要：

换热网络规模越大，其解空间内极值点呈指数性增长，优化时不仅要求算法具有强大的全局寻优能力，局部解空间的高精度搜索也不可或缺。鉴于强制进化随机游走算法（RWCE）优化换热网络时难以兼顾局部搜索能力、易导致遗漏最优解的现象，同时为增大算法优化大规模换热网络的种群数量，本文将精细搜索和并行计算相结合提出了并行双层RWCE算法。算法基于多核并行技术，通过并行线程分配建立基础层和精细层，基础层在并行计算技术加持下，算法全局搜索能力大幅提升，精细层将基础层当前最优解实时精细搜索，避免了原算法差解代替优解现象。最后通过两个算例进行验证，结果表明并行双层RWCE算法不仅具有更强的全局搜索能力，且兼具高精度的局部搜索能力，在优化进程中有效保护了最优解。

关键词: 过程系统, 换热网络, 优化, 算法, 双层算法, 并行计算