The Construction of the Comfort Model of Interior Space Environment Design Prediction Based on Neural Network
Through the calculation of environmental comfort index,five PMV variable indexes that affect the comfort degree of indoor space environment design are obtained.They are taken as the in-put of RBF neural network model after the weight threshold is improved by particle swarm optimization algorithm.After PMV transformation is completed by normalized deformation method,combined with the comfort evaluation standard.The final comfort level of indoor space environment design is obtained.Through the test and analysis,it is concluded that the fitting degree of the comfort prediction output value and the expected output value of the model is very high,and the average prediction error is only 0.21％,and it can accurately predict the comfort degree of the interior space environment design at differ-ent times.
万方数据
维普
