改进并行粒子群算法用于冷却水系统节能优化

Improved parallel particle swarm algorithm for energy-saving optimization of cooling water system

针对冷却水系统优化问题提出一种改进并行粒子群(IPPSO)算法,以系统能耗最小为优化目标,以系统中各设备的运行参数为优化变量进行求解.在该算法中,采用随机和混沌序列机制分别对两个种群的粒子进行初始化,使两种群在产生初期便具有不同特征;并根据两种群特点,采用不同惯性权重改进策略,提高算法搜索能力;同时利用一种新迁移算子对种群间个体进行交换,增强粒子多样性,使种群向更高层次进化;此外,考虑到系统设备运行数量为整数且受到系统设计总台数的限制,引入穷举法机制对系统中设备部分运行参数进行求解,减少最优解验证工作量,缩短优化时间.最后对某实际冷却水系统进行了详细测试,结果表明,使用IPPSO算法对设备运行参数优化后,冷却水系统总能耗降低12.49%,具有较好的节能效果.同时相比于其他算法,IPPSO能得到更好的优化策略,且在收敛性、计算复杂度和鲁棒性方面具有优势.

Aiming at the optimization problem of cooling water system,an improved parallel particle swarm optimization(IPPSO)algorithm is proposed.The optimization goal is to minimize the energy consumption of the system,and the equipment operating parameters in the system are taken as optimization variables to solve the problem.In this algorithm,the random and chaotic sequence mechanisms are used to initialize the particles of the two groups respectively,so that the two groups present different characteristics at the initial stage of generation;based on the characteristics of the two groups,different inertia weight improved strategies are adopted to improve algorithm search ability;at the same time,a new migration operator is proposed to exchange individuals between the groups to enhance the diversity of particles.In addition,because the number of system equipment running is an integer and limited by its total equipment number,an exhaustive method is introduced.Some equipment operating parameters in the system are optimized,reducing the workload of optimal solution verification and shortening the optimization time.Finally,a detailed test is carried out on an actual cooling water system,and the results show that after using the IPPSO algorithm to optimize the equipment operating parameters,the total energy consumption of the cooling water system is reduced by 12.49%,which performs a good energy-saving effect.Simultaneously,compared with the other algorithms,IPPSO can obtain better optimization strategies,and has advantages in convergence,computational complexity and robustness.