The traditional cellular automata has some problems such as insufficient consideration of dynamic elements, single division of cell types and neighborhoods, lack of scientific definition of transformation rules, and low model accuracy. This paper is based on the application of cellular automata in the field of urban planning, summarizes the optimization types of traditional cellular automata by domestic and foreign scholars, including the introduction of cell constraints, optimization of cells and neighborhoods, acquisition of transformation rules and application of space technology. Through the status quo and comparative research of the optimization types, proposes the direction that researchers need to pay attention to in the future, that is, introducing multiple type cells and neighborhoods, applying new gen-eration information technology, and expanding the application scope of CA, in order to further improve the accuracy of CA model and its universality and scientific application in urban planning.