基于概率神经网络(PNN)的社区居家养老建筑评价方法  

An application study of probabilistic neural network(PNN)on home-based care buildings for elderly in the community

作　　者：吴农[1] 吴蔚[2] Wu Nong;Wu Wei(School of Mechanics,Civil Engineering and Architecture,Northwestern Polytechnical University,Xi’an 710000,China;School of Architecture and Urban Planning,Nanjing University,Nanjing 210093,China)

机构地区：[1]西北工业大学力学与土木建筑学院,陕西西安710000 [2]南京大学建筑与城市规划学院,江苏南京210093

出　　处：《山西建筑》2021年第2期14-16,共3页Shanxi Architecture

基　　金：江苏省住房与建设厅科技指导类项目(项目编号:2020ZD55)。

摘　　要：近些年来我国一直在大力倡导和推进居家养老和社区养老的发展,然而,由于影响社区居家养老评价的各种主客观因素较多,目前的整体评估方法和体系都难以取得较为令人满意的结果。利用概率神经网络的结构和学习规则特性,对城市社区居家型养老建筑设施及其所处环境等级进行非线性分类和评价。在利用全国12个城市35个社区的样本数据进行训练后,对某城市5个社区居家型养老建筑设施及其所处环境等级进行仿真评价,来验证以人工神经网络为代表的人工智能评价模型在城市社区居家养老建筑设施及其所处环境等级评价中的实用性与可靠性。In recently year,China has been promoting home-based care for elderly in the community.However,there are many subjective and objective factors which affect the total evaluation for this home-based care model.This study is to apply Probabilistic Neural Networks(PNN)to study structure and non-linear category of elderly care model.Based on a training with the sample data of 35 communities in 12 cities across China,5 communities of home-based care for elderly are evaluated by using PNN,to verify its usability and reliability.

关 键 词：社区居家型养老建筑 概率神经网络 建筑环境评价 非线性分析 

分 类 号：TU241.93[建筑科学—建筑设计及理论]