Compromise Rank Genetic Programming for Automated Nonlinear Design of Disaster Management

This paper presents a novel multiobjective evolutionary algorithm, called compromise rank genetic programming (CRGP), to realize a nonlinear system design (NSD) for disaster management automatically. This NSD issue is formulated here as a multiobjective optimization problem (MOP) that needs to optimize model performance and model structure simultaneously. CRGP combines decision making with the optimization process to get the final global solution in a single run. This algorithm adopts a new rank approach incorporating the subjective information to guide the search, which ranks individuals according to the compromise distance of their mapping vectors in the objective space. We prove here that the proposed approach can converge to the global optimum under certain constraints. To illustrate the practicality of CRGP, finally it is applied to a postearthquake reconstruction management problem. Experimental results show that CRGP is effective in exploring the unknown nonlinear systems among huge datasets, which is beneficial to assist the postearthquake renewal with high accuracy and efficiency. The proposed method is found to have a superior performance in obtaining a satisfied model structure compared to other related methods to address the disaster management problem.