An integrated computational intelligence framework for construction performance diagnosis.

摘要

Construction performance diagnosis (CPD), the process of finding and explaining performance problems, is a vital part of the project control process. Generally in construction, a diagnostic problem arises if there is a discrepancy between the actual performance of resource(s) and the planned performance. The diagnostic task is to determine the cause(s) of this discrepancy. Understanding what caused an event to occur enables the construction manager to predict, to plan for, to prevent, and to explain the occurrence of the event. Automating the performance diagnosis process to detect, diagnose, and report results within a time frame that permits prompt field response can significantly enhance the project control process.; This thesis investigates the advantages of introducing computational intelligence tools to develop automated performance diagnostic models to explain construction performance. The integrated diagnostic system has advantages of both fuzzy systems (e.g., the use of expert knowledge representation and the ability of explaining generated decisions) and neural-network systems (e.g., ability of learning, adaptation, optimization, and high fault tolerance). Additionally, the powerful globaloptimization technique of genetic algorithms effectively optimizes the network structure to provide the best solution.; In this thesis, several key issues and challenges of developing robust performance diagnostic models for construction-related problems are discussed. The essential features of the model are described in detail. The efficiency and effectiveness of the techniques and methods developed in this thesis are tested in the domain of industrial construction labor productivity and implemented in a computer system called XCOPE.; The main contributions of this work are twofold. One contribution is the development of a unified integrated computationally intelligent framework to diagnose construction performance. Another contribution is in the acquisition and representation of a construction expert's knowledge. Several different techniques, such as Nominal Group Technique (NGT), Semantic Differential (SD) Approach, and Fuzzy Membership Functions, are explored to select the most suitable knowledge acquisition and representation techniques for construction performance modeling.