室内环境舒适度的神经网络建模与仿真

摘要

The main purpose of air conditioning system is to provide a comfortable enviroment, the relationship between the thermal comfort index SET * and those environmental factors which affecte SET * is very complicated and nonlinear, therefore it can not get the real -time thermal comfort index, and also it can not meet the requirements of real - time control. In the paper, the real - time thermal comfort index SET * can be obtained by BP neural network using optimized L - M algorithm. With the SET * index as output, and the environmental factors as inputs,the neural network prediction model of SET * index is established. The results of simulation show good agreement between the thermal comfort index SET * calculated from the neural network model in real - time and those calculated from iterative formula, which ensures the validity of the model.%空调控制系统是提供给人一个舒适的热环境,影响热环境的舒适度指标SET*值与影响它的环境因素之间具有复杂性和非线性等特点,针对能实时的确定人体舒适度,为了能够满足空调系统实时控制的要求.采用优化L-M算法的BP神经网络方法能够控制实时的确定SET*指标,分析热环境因素与SET*指标的关系,以SET*指标作为输出,影响SET*的环境变量作为输入,建立了SET*指标的神经网络模型.仿真结果表明神经网络模型实时计算的SET.值与迭代计算得到的SET*值相一致,保证了室内舒适度的效果.