Abstract

In recent years, the rapid development of the world’s economy has led to the large-scale development and utilization of ecological resources on the earth, due to which the ecological environment has been continuously and seriously damaged, resulting in the waste of resources, soil erosion, land desertification, etc. To avoid further damage to the ecological environment and ecological resources, improve the utilization rate of ecological resources, and ensure the sustainable development of human society, it is necessary to evaluate the ecological environment. In this study, we collected the required data using the Delphi method and remote sensing technology. Secondly, the green Olympic building evaluation system (which refers to the CASBEE method in Japan) was used to evaluate the impact of green roofs on architectural design and the urban ecological environment. Third, a deep learning (DL)-based hybrid model, which consists of a convolutional neural network (CNN) and long–short-term memory (SLSTM), known as CNN–LSTM, was used to evaluate the impact of green roofs on urban ecology and building architectural design. The influence of thermal comfort on the indoor environment of green roof buildings was studied. For experimentation, six samples of Shanghai Thumb Plaza, Splendid Tesco Point, Chaoshan Yuan Hotel, Green Management Office, Huangpu District Domestic Waste Transfer Station, and Changning District Fuxin Slaughterhouse were selected as evaluation objects, and the effect of green roofs on building design and urban ecology was evaluated from six levels: ecological, ornamental, safety, functional, social, and economic. Both the CASBEE and DL-based methods, CNN–LSTM, performed well and increased the evaluation results to some extent. The CNN–LSTM model increased the accuracy of the system by 3.55%, precision by 3.50%, recall by 4.46%, and F1-score by 3.30%. Overall, this study summarizes the existing problems of green rooftop buildings in Shanghai at this stage, which is conducive to formulating optimization strategies to improve the ecological benefits of green roof buildings and has important practical significance for realizing the sustainable development of human society.

中文翻译：

---

使用深度学习方法的绿色屋顶及其对建筑设计和城市生态的影响

摘要

近年来，世界经济的快速发展导致地球生态资源的大规模开发利用，导致生态环境不断受到严重破坏，造成资源浪费、水土流失等。为避免生态环境和生态资源进一步破坏，提高生态资源利用率，保障人类社会可持续发展，有必要对生态环境进行评价。在本研究中，我们利用德尔菲法和遥感技术收集了所需的数据。其次，采用绿色奥运建筑评价体系（指日本CASBEE方法）评价绿色屋顶对建筑设计和城市生态环境的影响。第三，基于深度学习（DL）的混合模型，由卷积神经网络（CNN）和长短期记忆（SLSTM）组成，称为CNN-LSTM，用于评估绿色屋顶对环境的影响。城市生态与建筑设计。研究了热舒适度对绿色屋顶建筑室内环境的影响。实验选取上海大拇指广场、锦绣乐购点、潮汕园酒店、绿化管理处、黄浦区生活垃圾中转站、长宁区阜新屠宰场6个样板作为评价对象，研究屋顶绿化对建筑设计和施工的影响。城市生态从生态、观赏、安全、功能、社会、经济六个层面进行评价。CASBEE 和基于 DL 的方法 CNN-LSTM 都表现良好，并在一定程度上提高了评估结果。CNN-LSTM模型将系统的准确率提高了3.55%，精确率提高了3.50%，召回率提高了4.46%，F1-score提高了3.30%。总体而言，本研究总结了现阶段上海绿色屋顶建筑存在的问题，有利于制定提高绿色屋顶建筑生态效益的优化策略，对实现人类社会的可持续发展具有重要的现实意义。