Simulation of land use change using genetic algorithms neurology network based cellular automata

Min Cao,Xiao Shi,Shanshan Tan

DOI: https://doi.org/10.1109/ICMULT.2010.5629656

2010-01-01

Abstract:Cellular automata(CA) have been increasingly used to simulate complex land use systems. Empirical data can be used to calibrate cellular automata models. Thus, realistic urban patterns can be generated. Traditionally, artificial neurology network(ANN) was used to calibrate cellular automata models. As artificial neurology network easily fall into local minimum value, the genetic algorithms neurology network based cellular automata model (GANN-CA) is developed by using object-oriented modeling idea and combining with genetic algorithm neural network and cellular automata. Taking the north branch of Yangtze River estuary as an example, its land use change model is constructed and applied to the simulation and prediction of this land use change, which supply a good foundation of the land use Plan. The Research demonstrates that the GANN-CA model achieves a better simulation effect. It is concluded that GANN-CA model can make full use of artificial neural network to obtain the variable space parameters and simplify the land use transfer rule. This model considered more comprehensive space impact factors and optimized the connection weights and thresholds of neural network, which improved and expanded the artificial neurology network based cellular automata model (ANN-CA). ©2010 IEEE.