Soft computing approach to construction performance prediction and diagnosis

Increasing project complexity and greater levels of fast-tracking on construction projects make the need for quickly detecting and diagnosing a deviation in a construction performance measure from its planned value a challenging task. In such a rapidly changing environment, timely detection of deviations is critical so that the most effective corrective actions can be taken. This paper presents an integrated model that is capable of predicting and diagnosing construction performance deviations based on a combination of field measurements and subjective assessments of performance-related factors. The proposed system is based on the synergistic integration of the soft computing approaches of fuzzy set theory, neural networks, and genetic algorithms. A systematic methodology to elicit and represent qualitative construction performance knowledge from a group of experts is presented. The essential features of the model are described in detail and are implemented in a computer system called XCOPE (explaining construction performance).