Simulating the Construction Process Using Neural Networks

The paper describes the development of an efficient method of simulation construction activity using artificial neural networks. A brief introduction to construction simulation is provided, and a major limitation of the technique, that of unaccepatably lengthy processing periods, is highlighted. This is followed by a description of the main mechanisms if neural networks and their potential as a means of increasing the execution speed of construction simulations. A modular neural network approach is proposed as the basis of a simulation facility suitable for general application to construction. The main types of modules required are identified, and the ways in which they can be implemented in neural circuitry are explored. The paper concludes with a general evaluation of the performance of the system, illustrating that it could operate many orders of magnitude faster than existing computer based simulation facilities.