换热站并联水泵分布式优化控制

针对现有换热站并联水泵优化算法在集中式架构下控制适应性不足的问题,本文提出了一种改进的分布式并联水泵优化算法.首先,建立了并联水泵的分布式控制系统,并对该优化问题的数学模型进行描述,在目标函数中引入自适应非线性因子;然后,设计了改进的分布式果蝇优化算法,在该算法中每台水泵的控制器仅通过与邻居控制器交互信息即可完成并联水泵的优化;并且,在嗅觉搜索阶段,使用正弦余弦策略替代赋予个体距离与方向的随机策略;最后,以两个实际换热站中不同并联水泵系统为例对算法进行仿真验证,并基于仿真结果进行性能分析.结果表明,相较于传统算法,改进的分布式果蝇优化算法能得到更优的控制策略,有着收敛速度快、稳定性好和鲁棒性强的特点;并且该算法适用于不同系统的并联水泵优化问题,具有可扩展性.在实际工程验证中相较于集中式算法,该算法在总功率和计算时间上分别平均降低了5.47%和29.90%,因此,能够满足实际换热站中对并联水泵热负荷优化分配的需求.

Distributed optimal control of parallel water pumps in heat exchange station

Aiming at the problem that the optimization algorithm of parallel circulating pump in existing heat exchange station has insufficient control adaptability under centralized structure, an improved distributed parallel circulating pump optimization algorithm was proposed. Firstly, a distributed control system of parallel circulating pumps was established, and the mathematical model of the optimization problem was described, and the adaptive nonlinear factor was introduced into the objective function. Then, an improved distributed fruit fly optimization algorithm was designed, in which the controller of each water pump can complete the optimization of parallel circulating pumps only by interacting with adjacent controllers. In the olfactory search stage, the sinusoidal cosine strategy was used to replace the random strategy given individual distance and direction. Finally, the algorithm was simulated and verified by different parallel circulating pump systems in two actual heat exchange stations, and the performance was analyzed based on the simulation results. The results show that compared with the traditional algorithm, the improved distributed fruit fly optimization algorithm can obtain a better control strategy, with the characteristics of fast convergence, good stability and strong robustness. The algorithm can be applied to the parallel pump optimization problems of different systems, and has scalability. Compared with the centralized algorithm in the actual engineering verification, the total power and computing time of the proposed algorithm are reduced by 5.47% and 29.90%, respectively. Therefore, it can meet the demand for optimal distribution of the heat load of parallel water pumps in actual heat exchange stations.