基于神经网络与NSGA-Ⅱ算法的方舱温控优化设计

Optimization Design of Shelter Temperature Control based on Neural Network and NSGA-ⅡAlgorithm

综合考虑车载方舱外部传热与舱内设备的功率负载等因素对舱内热环境的影响,以车载空调的布局与送风角等作为设计变量,以舱内电子设备耐受温度与乘员舒适度为目标函数,采用BP神经网络建立设计变量与目标的映射关系,利用改进型非支配排序遗传算法对舱内的气流布局与环境温度梯度进行优化,取得Pareto前沿解。优化结果表明,在制冷功率一定的前提下,最优解集主要集中在空调较接近电子设备且前出风角较平直时。此方法节省仿真优化成本,对方舱内温度与流场环境控制具有重要工程指导意义。

Considering the thermal environment influence of heat transfer outside the shelter and the power load of the electronic equipment in the shelter,we took the layout of the car air conditioner and the air supply angle as design variables,and took the withstand temperature of electronic equipment in the shelter and occupant comfort as the objective function,and BP neural network mapping relationship between variables and objective was established.By using no-dominated sorting genetic algorithm,the optimization of airflow layout and ambient temperature gradient were done,and we achieved Pareto front solution.The optimization results showed that under the premise of constant cooling power,the optimal solution set was mainly concentrated in the air conditioner close to the electronic equipment.This method saved the simulation optimization cost,and had important engineering guiding significance for the temperature and flow field environment control in the shelter.