Application of Projection Pursuit Evaluation Model Based on Real-Coded Accelerating Genetic Algorithm in Evaluating Wetland Soil Quality Variations in the Sanjiang Plain, China

摘要

A new technique of dimension reduction named projection pursuit is applied to model and evaluate wetland soil quality variations in the Sanjiang Plain, Helongjiang Province, China. By adopting the improved real-coded accelerating genetic algorithm (RAGA), the projection direction is optimized and multidimensional indexes are converted into low-dimensional space. Classification of wetland soils and evaluation of wetland soil quality variations are realized by pursuing optimum projection direction and projection function value. Therefore, by adopting this new method, any possible human interference can be avoided and sound results can be achieved in researching quality changes and classification of wetland soils.