采用改进被囊群算法的多冷水机组负荷分配优化

Optimal Load Allocation of Multiple Chiller System Using Improved Tunicate Swarm Algorithm

摘要:

为了降低中央空调系统的运行能耗，针对多冷水机组负荷分配优化问题，提出一种随机森林特征优选结合核函数极限学习机的冷水机组能效预测模型，通过剔除冗余特征提高预测精度；然后提出一种混合策略改进的被囊群算法，融合鲸鱼螺旋搜索策略改进个体更新方式，引入非线性动态权重平衡全局探索和局部开发，使用空翻扰动策略避免陷入局部最优；最后在能效模型的基础上，采用改进被囊群算法对多冷水机组负荷分配进行优化。实验结果表明，随机森林特征优选的方法可以有效的提高能效预测模型的准确度；改进被囊群算法通过优化机组的启停状态和负荷率可以有效发挥系统的节能潜力，与原有方法相比能耗降低约6%。说明该方法适用于多冷水机组的负荷分配优化问题。

Abstract:

In order to reduce the energy consumption of the central air-conditioning system, and aim at the problem of the optimal load allocation in multiple chiller system, a prediction model of energy efficiency in chillers is proposed. Based on random forest feature optimization combined with kernel function extreme learning machine, the model improves the prediction accuracy by eliminating redundant features. Then an improved tunicate swarm algorithm based on hybrid strategy(ITSA)is proposed. Firstly, whale spiral search strategy is coalesced to improve individual update methods. Secondly a non-linear dynamic weights is introduced to balance global exploration and local development. Thirdly somersault strategies are used to avoid falling into partial optimal. Finally, on the basis of the energy-efficiency model, ITSA is used to optimize load allocation of multiple chiller system. The experimental results show that the random forest feature optimization can effectively improve the accuracy of the energy efficiency prediction model. ITSA can effectively exert the energy saving potential of the system by optimizing the on-off status and load ratio of the chillers. Compared with the original method, the energy consumption can be reduced by about 6%, which shows that the method is appropriate for optimal load allocation of multiple chiller system.