Prediction Model for Soil Heavy Metal Content Based on COPSO-GRNN
The prediction of soil heavy metal content is an important part of soil pollution control.To improve the accuracy of prediction,this paper proposes a prediction model for soil heavy metal content based on COPSO-GRNN.In response to the problem that it has difficulty in determining the smoothing factor of generalized regression neural networks(GRNN),the model uses cosine optimization particle swarm optimization(COPSO)for optimization.In addition to adding a small population comparison strategy to the population,it also uses cosine acceleration coefficient to expand the search range and avoid falling into local optima during the optimization process.Then,an adaptation criterion is introduced to improve the convergence speed of the algorithm.Comparative experiments are conducted between this model and several common prediction models for soil heavy metal content.The experimental results show that the predicted values of this model are closer to the true values and has better predictive performance.
the prediction of soil heavy metal contentGRNNCOPSOparameter optimization
国家科技期刊平台
NSTL
万方数据
