基于ARIMA和BP神经网络的生态系统服务价值预测仿真

针对现有生态系统服务价值预测体系误差大、研究不足等问题,本文构建了面向未来生态服务价值的预测机制。该机制采用主成分分析法和Person相关系数法,建立了以GDP、人口、土地利用率、土地管理政策和城市人口密度为驱动力的生态系统服务价值关联矩阵,提出了基于ARIMA和BP神经网络的生态系统服务价值的时间序列预测机制。为验证预测机制有效性,本文选取中国主要土地利用类型代表省份的真实土地数据集进行分析,研究结果表明本文建立的预测模型平均绝对误差仅为0.023,且从预测结果来看,未来草地生态会向较好趋势发展,林地生态发展不容乐观。

Simulation of Ecosystem Service Value Prediction Based on ARIMA and BP Neural Network

Aiming at the problems of large error and insufficient research in the existing ecosystem service value prediction system,this paper establishes a prediction mechanism for future ecosystem service value.This mechanism adopts principal component analysis and Person correlation coefficient method,1establishes the correlation matrix of ecosystem service value driven by GDP,population,land use rate,land management policy and urban population density,and constructs a time series prediction mechanism of ecosystem service value based on ARIMA and BP neural network.In order to verify the validity of the prediction mechanism,the real land data sets of representative provinces of China’s main land use types are selected for analysis.The results show that the average absolute error of the prediction model is only 0.023,and from the future prediction results,the grassland ecology will develop towards a better trend and the forest ecology will develop in the future negatively.