Research On The Inverse Design Of Indoor Environment Based On Intelligence Algorithm And Fuzzy Logic Control

The inverse design method on indoor environment is a hot topic.It can obtain the values of design variables according to the requirement of design objectives.It provides theoretical guidance to the design of indoor environment and has widely commercial prospects.In this paper,a high-efficiency inverse design method for indoor environment is proposed based on the comprehensive analysis of existing methods.It combined the genetic algorithm(GA),artificial neural network(ANN)and fuzzy logic control(FLC)based on computational fluid dynamic(CFD).The Blay model and aircraft cabin model are used to test the inverse design method.When the Blay model is used,the design variables are the inlet velocity and temperature,and the velocity and temperature at the monitoring point are designed.When the aircraft cabin model is used,the design variables are the inlet velocity,temperature,angle,location and outlet location,and the thermal comfort,draught rate,air age,local velocity,and vertical temperature difference are designed.The results are as follows:The inverse design method based on GA is studied first.The single-objective GA is used to the inverse design of Blay model and the muti-objectives GA is used to the inverse design of cabin model.Non-domination method is used to solve the sorting problem in the muti-objectives design.The hypervolume is used to obtain the interval solutions of inverse design,and 8 interval solutions are obtained.The optimal interval solution is selected by the value of hypervolume.Large computational cost is needed when only GA is used as the engine of inverse design.To reduce the computational cost,the ANN is involved in the inverse design method.ANN is used to predict the design objectives of new individuals generated by GA.To eliminate the design error induced by ANN,both ANN and CFD are used to obtain the design objectives of new individuals.The logarithm normalization method and the optimization of initial generation are proposed to improve the efficiency of inverse design.A self-updating ANN is then proposed to automaticly determine the number of training samples for ANN.With the help of ANN,the computational costs for the inverse design of Blay and cabin model are reduced by 49.4% and 60.8%,separately.Then the use of FLC in the inverse design is studied.The computing probability is obtained when FLC and ANN are combined.The evolution space is obtained when FLC and GA are combined.The computational cost for the inverse design of 2D and 3D cabin model are reduced by 26.6% and 24.6%,separately.In conclusion,the efficiency of the inverse design of indoor environment is improved by the proposed inverse design method.The computational cost for the inverse design of cabin model is reduced by 71.3%.