混沌神经网络住宅性能综合评价模型的研究

Synthetic evaluation to quality and functions of residential building using chaos artificial neural network

住宅性能综合评价是住宅商品化和市场化的重要基础性工作.利用人工神经网络模拟人脑学习和记忆功能的特点,建立了住宅性能综合评价模型.针对神经网络BP算法收敛速度慢容易陷入局部极小值问题,将混沌理论引入神经网络算法中,利用混沌运动的遍历特性加快了训练速度,使其收敛于全局最优解.算例表明,混沌神经网络模型可以准确地评价住宅的综合性能,有良好的应用前景.

It is important to evaluate a residential building synthetically in the building market, the prices of houses are greatly relied on the their quality and functions. A method of synthetic evaluation to residential building is presented. This method is based on chaos artificial neural network (CANN), which imitates the function of learning and memory of human being. The motion of chaos is applied in the learning (or train) of BP algorithm of ANN to improve its convergence. This makes its solution escape from the local utmost points and approach global optimum. The results of simulations show that the CANN method can synthetically evaluate the quality and function of a residential building and will be a good candidate for the scheme of building evaluation in the future.