Big Data and Machine Learning-enabled Automated BIM for Projects (Auto-BIM): A Common Data Collaborative System for Improved Project Performance

"BIM is touted as an effective way of addressing issues affecting the productivity of the construction industry. The task-force on BIM hypothesises that with BIM-adoption, ""significant improvement in cost, value and carbon-performance can be achieved through the use of open-sharable asset-information"". To reinforce its benefits, the government Construction-2025 lists BIM as a key-element for achieving its goal of 33% lower-cost, 50% faster-delivery, 50% lower-emissions and 50% improvement in export.

Although there has been an increase in BIM-adoption, companies still find it difficult to implement the ""real"" BIM and realise the expected benefits. This is because of the naming convention in line with PAS-1192 and the need for adequate building-information to accompany 3D-representation of building materials/elements/products in a collaborative environment. For organisations that have surpassed the barrier to BIM-adoption, the main-challenge remains getting everyone involved in collaborative-projects to use CDE and to ascertain the exact-level of(and the specific) information required for different aspects and types of assets. Thus, some projects on which BIM is claimed to be used have only assembled digital information without providing useful information for construction, in the short-term, and data for asset-management in the long-term.

Notwithstanding these challenges, there is currently no tool to support organisational BIM-adoption and compliance with the standard, leverage previous project lessons/historic data, support automated Construction-Operations-Building-Information-Exchange-(COBie) and facilitate supply-chain integration with product-manufacturers. Based on these, the project adopts techniques in Machine-Learning-(ML) and Big-Data-Analytics to create an innovative tool-(Auto-BIM) as a plug-in to BIM-tools. It consists of four-elements as follows:

1\.**Automated-Naming-of-BIM-model-in-a-CDE-approach(Auto-BIMName)--**This helps project team to name their files in consistency/compliance with PAS-1192 and BS-EN-ISO-19650\. It would also help in automatically mapping the title-block, which is currently being done manually between collaborating companies/originators and roles.

2\.**Automated-Population-of-Building-Information(Auto-BIMPopulate)--**This will prepopulate the 3D-representation of products/elements with relevant metadata including the Omniclass classification, model number, service information, materials, etc. This will facilitate a conventional approach to project communication/collaboration, and accelerate BIM-adoption and benefit-realisation.

3\. **Automated-Sharing-of-BIM-Objects-and-Model-Data****(Auto-BIMShare)--**The Auto-BIMShare provides a unique platform for sharing reusable object library and associated information to facilitate common-language across software boundaries. It will also provide opportunities for manufacturers to make their products/materials available for potential specifiers and buyers. The Auto-BIMShare would facilitate co-creation/sharing of information between the design, procurement, and maintenance/operation team within/across projects

4\.**Automated-BIM-learning-Platform(Auto-BIMLearn)--**BIM currently has no capacity for diagnosing projects. The Auto-BIMLearn would leverage on historical data, tacit knowledge(lesson-learnt) and asset management-information to support design, construction and asset management decisions."