换热网络优化中换热单元生成频次的影响分析及策略改进

换热单元的生成频次对换热网络优化有直接的影响，分析其机理可以指导求解算法的改进。本文采用节点非结构模型及强制进化随机游走算法，通过设置不同的换热单元生成概率与生成个数以改变换热单元的生成频次，观察并记录个体在优化过程中产生的年综合费用变化情况。发现不同生成频次下整型变量和连续变量的相对优化频率发生了变化，从而改变了整体优化过程，并且与个体当前优化状态不适应的生成频次会造成相对优化频率不平衡、优化结果较差。基于此，本文提出了一种具有生成参数动态调节策略的换热单元生成方式，在优化过程中根据个体状态实时调整换热单元的生成概率与生成个数，平衡整型变量与连续变量的相对优化频率。最后，采用15SP、10SP和20SP算例进行验证，换热网络的年综合费用较文献结果分别下降了1.06%、0.16%、0.68%，验证了该策略使得换热单元的生成频次更为合理，有效地提高了算法的优化效率。

Influence analysis and strategy improvement of heat exchanger generation frequency in heat exchanger networks optimization

The generation frequency of heat exchanger directly affects the optimization process of heat exchanger networks (HENs). The analysis of its mechanism can guide the improvement of the optimization algorithm. The Node-Wise Non-Structural Superstructure model and Random Walk Algorithm with Compulsive Evolution algorithm were adopted. Different generation probabilities and numbers of heat exchanger were set to change the generation frequency of heat exchanger. Then, the change of total annual cost (TAC) which generated by individuals in the optimization process was observed and recorded. The analysis showed that the generation frequency of heat exchanger can change the whole optimization process by affecting the relative optimization frequency of integer variable and continuous variable. And the generation frequency which is not suitable for the current state of the individual will cause the relative optimization frequency imbalance, and the optimization results are poor. Based on this, a generation pattern of heat exchanger with dynamic adjustment strategy of generation parameters was proposed, which is to change the individuals' generation probability and number in the optimization process. Then, the relative optimization frequency of integer variable and continuous variable was adjusted to make it more suitable for individual optimization. The generation probability and the number of heat exchanger were adjusted in real time according to the individual state in the optimization. And the relative optimization frequency of integer variables and continuous variables was balanced. Finally, the example of 15SP， 10SP and 20SP was used for verification. The TAC of HENs respectively saved 1.06%, 0.16% and 0.68% as compared to the optimums in literature. The results showed that the strategy effectively improves the optimization performance of the algorithm and makes the generation of heat exchanger more reasonable.