Leveraging operational data for intelligent decision support in construction equipment management

Construction equipment management is aimed at managing equipment resources in order to maximize return on capital investments and satisfy the needs of project management in a timely and cost-effective manner. A rapid development of computer software and hardware, along with various automation methods for data acquisition, has catalyzed the computerization of construction equipment management in recent years. The primary objective of this research is to investigate the application of cutting edge data warehousing and data mining technologies for intelligent decision-making in construction equipment management. In cooperation with Standard General Inc. , a large road building and maintenance contractor in Alberta, Canada, and based on the M-Track equipment information management system developed by NSERC/Alberta Construction Industry Research Chair as well as nine years of equipment operational data, this research proposes to improve the M-Track system in its data analysis and decision support capabilities using advanced computer tools. As a response to the problems existent in Standard General Inc. , and common to the construction industry in general, the research addresses issues of (i) How to design and implement an information infrastructure which advocates data sharing, information retrieval, and knowledge discovery for fact-based intelligent construction equipment management; (ii) how to apply the data warehousing technique in construction equipment management for decision support; (iii) how to leverage the large amounts of operational data for automated knowledge generation and decision support using data mining techniques. A number of tests and demonstrations on the prototype applications have proven that data warehousing and data mining are suitable technologies for improving the current practice of decision support through an integrated data repository, need-based information retrieval and decision analysis, and automatic identification of trends, patterns, or rules from data. The novel, non-parametric outlier mining algorithm developed in this research has proved to be effective and efficient in detecting the top-N most interesting outliers from equipment database. Although developed in the field of construction equipment management, the documented findings, proposed methods, and recommended best practice are equally applicable to other areas of construction management.