Method for optimizing land use evolution CA model transformation rules

Abstract

The invention provides a method for optimizing land use evolution CA model transformation rules, and the method comprises the following steps: firstly, according to the multitemporal spatiotemporal land use data of a studied area, selecting factors influencing the land use change of the studied area as the spatial characteristic variables of land use evolution CA; establishing a Logistic regression method based land use evolution CA model, selecting sampling point data, and calculating the influence weight of each space variable; and optimizing the parameters of the land use evolution CA model and establishing an evolution programming algorithm based land use evolution CA model by using an evolution programming method, and finally, verifying the simulation accuracy and efficiency of the model. Compared with the existing commonly-used method for obtaining land use evolution CA model transformation rules, the method provided by the invention is simpler in operation, faster in convergence speed and higher in simulation precision; and compared with other methods for obtaining land use evolution CA model transformation rules, the method provided by the invention is more identical with the essential characteristics of spatiotemporal land use evolution.

Description

technical field [0001] The invention belongs to the technical field of geographic information science and technology, and particularly relates to introducing an evolutionary planning algorithm to optimize the conversion rules of a massive space-time land use evolution CA simulation model. Background technique [0002] Cellular Automata (CA) is a dynamic system of micro-individual interaction, time and space discretization. It was first proposed by American mathematicians Ulam and Neumann in the 1940s, and later by British computational scientist Wolfram. In the 1980s, its definition was given and applied to complex system simulation. CA has powerful spatial computing capabilities and is often used in the study of the evolution process of self-organizing systems. It is a grid dynamics model with discrete time, space and state, and local interaction and temporal causality. The ability of the system to evolve over time. Its "bottom-up" research idea fully embodies the idea th...

AI Technical Summary

Problems solved by technology

[0005] Analytical hierarchy process (AHP) and multi-criteria discriminant (MCE) are used to determine the optimal parameter combination scheme of land use evolution CA, but this method cannot eliminate the negative impact of multi-collinearity of spatial variables, so it is difficult to accurately simulate land use evolutionary phenomenon

[0006] Using the decision tree method to determine the conversion rules represented by the See5.0 decision tree model is easy to fall into a local optimal situation

The method of using the nuclear learning machine has the problems of unclear physical meaning of the conversion rules and a large amount of calculation. The neural network method is also used to automatically obtain the conversion rules. Although it is very convenient to determine the model parameters and model structure, it eliminates the problems caused by conventional methods. However, there are also problems such as unclear physical meaning of conversion rules and slow convergence speed.

[0007] Using the Logistic regression method to determine the conversion rules has a relatively clear physical meaning, but it is too simple to extract complex physical phenomena using a linear method. Therefore, it is necessary to optimize the parameters obtained based on the Logistic regression method, such as using genetic Algorithm optimization to obtain conversion rules, on the one hand, can retain the physical meaning of the original conversion rules, and on the other hand, can also improve the accuracy of model simulation, but it faces problems such as complex operation and large transportation volume

At the same time, the spatio-temporal land use change data has the characteristics of massive and high-dimensional, which will inevitably increase the search space of the algorithm and increase the blindness of the search.

Images