Keywords: ANNs, Lagrange, optimization, construction planning, construction scheduling, big data in construction.

One of the biggest obstacles in optimization is finding explicit objective functions that describe target outcomes. However, the need to explicitly define complex objective functions and constraints with respect to design variables can be eliminated through the use of Artificial Neural Network (ANN)-based optimization. This method enables the optimization of discontinuous, nonlinear design problems with multiple variables, objective functions, and constraints. In this study, a scheduling simulation is established to generate big data for construction planning of a warehouse project. The complex construction process scheduling is formulated into an objective function derived from ANNs, which is trained on the generated data to map zoning areas and manpower to costs. Jacobian and Hessian matrices of the ANN-based functions are formulated to implement the Newton-Raphson iteration for finding stationary points of the Lagrange functions. The optimization can consider planning constraints such as site capacity, concreting capacity, etc., while providing solutions for minimizing costs. Results show that cost predicted by ANN-based optimization is located at the minimum of big data ranges, indicating a potential of the proposed method to aid construction engineers in establishing optimized construction strategies.

download the full-text of this paper (PDF, 12 pages, 1001 Kb)

go to the previous paper
go to the next paper
return to the table of contents
return to the volume description