BIM based Machine Learning framework for healthcare facilities

Traditional design approach for construction projects spans over several steps, from design phase, collaboration phase, construct phase and operation and management phase. Each phase consists of its own methodology and intermediate steps or intervals. Preliminary stages of a project involve data col...

ver descrição completa

Detalhes bibliográficos
Autor: Hogmo, Bjorn
Tipo de documento: dissertação
Data de publicação:2023
País:España
Recursos:Universitat Politècnica de Catalunya (UPC)
Repositório:UPCommons. Portal del coneixement obert de la UPC
Idioma:inglês
OAI Identifier:oai:upcommons.upc.edu:2117/385148
Acesso em linha:https://hdl.handle.net/2117/385148
Access Level:Acceso aberto
Palavra-chave:Health facilities -- Design and construction
Building information modeling
Particle Swarm Optimization
PSO
Hospital layout problem
HLP
BIM
Dynamo
Building information modelling
Equipaments sanitaris -- Disseny i construcció
Modelatge d'informació de construcció
Àrees temàtiques de la UPC::Edificació
Descrição
Resumo:Traditional design approach for construction projects spans over several steps, from design phase, collaboration phase, construct phase and operation and management phase. Each phase consists of its own methodology and intermediate steps or intervals. Preliminary stages of a project involve data collection and specification of the requirements, among several points, and within a healthcare facility the specifications are given by stakeholders and the users of the facility. To create good projects, the quality of the specifications relies heavily on the experience of the representants, while the interpretation of the specifications and creation of design rely on the experience of the designers. Such a manual creation of data collection, specifications and design are prone to human errors, only assessing a few alternatives and what worked earlier design approach. Contrary, machine learning (ML) and Building information modelling (BIM) software could assess thousands of alternatives and modified according to different optimization parameters. Therefore, the creation of a preliminary design model based on actual clinic data and their electronic health records (EHR), optimization of layout based on EHR and patients’ movement, which again could be connected to a automate creation of BIM model. Each of the separate areas, creating a machine learning on one side and atomate create BIM on the other hand has a potential benefit to generate savings, reduce length of preliminary phase and reduce labor hours. This thesis suggests a methodology for combining ML algorithms for optimization in hospital layout problems (HLP) design with the automate creation of 3D BIM model. Such a methodology is shown possible, and the report concludes such an approach is feasible. The methodology steps use of machine learning algorithm to create optimized layout solutions with coordinates in the 2-dimensional (2D) plane, where a visual programming Automa create a 3D BIM model for use in the preliminary phase