the research will be concerned with the development of a new and efficient approach to the simulation of construction processes, based on the use of artificial neural networks. The primary objective is to capitalize on the automatic modeling capabilities and inherent speed of neural networks, so as to make simulation a more accessible and productive tool for planners. Alternative schemes for implementing construction simulation in neural network environment will be evaluated to determine that with the greatest efficacy. This will be measured in terms of the speed of execution of a simulation, the ease with which valid models can be built, the accuracy of simulation results. The comparison will be made over a comprehensive range of modeling situations. consideration will be given to all principal types of neural architecture and training procedure that have potential application to the problem. The solution is expected to increase simulation speed by several orders of magnitude relative to conventional simulation methods, make processing speed independent of the complexity of a model, and facilitate the task of building valid models. Many of the concepts and ideas developed in the research could be extended to simulation applications outside the field of construction.
