| 摘要: |
| 【目的】准确评价陕甘宁新农业资源可持续利用水平。【方法】本文选取陕甘宁新地区为研究对象，从经济、科技、社会、自然环境、资源、生态治理6个方面构建了农业资源可持续利用的评价指标体系，并结合2018—2020年的相关数据，运用小波-PSOSVM评价农业资源可持续利用水平。【结果】①小波-PSOSVM农业资源可持续评价均方误差MSE为9.411 5×10-5，相关系数为0.968；而PSOSVM在同样的训练集以及同样的测试集下，得到的均方误差MSE、相关系数分别为0.015 3、0.967。小波处理后，PSOSVM预测的精度有所提高，收敛稍加快。②小波-SVM农业资源可持续评价均方误差MSE为20.836，相关系数为0.748；而SVM在同样的训练集以及同样的测试集下，均方误差MSE、相关系数分别为30.903、0.634，小波处理后，SVM预测的精度提高，收敛也稍快。【结论】PSO优化后的SVM，预测的精度提高较多，收敛也快很多。陕甘宁新地区农业资源可持续利用评价结果表明，新疆排名第一，甘肃第二，陕西第三，宁夏第四。 |
| 关键词: 小波分析；微粒群算法；支持向量机；资源可持续利用评价 |
| DOI：10.13522/j.cnki.ggps.2022222 |
| 分类号: |
| 基金项目: |
|
| Sustainable Agricultural Resource Utilization in Northwestern China Determined Using the Wavelet-PSOSVM |
|
WANG Chenxuan, CHEN Li, ZHANG An’an
|
|
1. School of Economics and Management, Xiamen University of Technology, Xiamen 361024, China;
2. School of Economics and Management, Anhui Jianzhu University, Hefei 230601, China
|
| Abstract: |
| 【Objective】The initiative of “Western Development” launched by Chinese Government in 1999 had greatly benefited the northwestern provinces in the country including Shaanxi, Gansu, Ningxia and Xinjiang. As ecological systems in these provinces are fragile, rational use of agricultural resources is crucial to their sustainable development. The purpose of this paper is to present a method to help sustainable use of their agricultural resources. 【Method】The method is based on the wavelet-PSOSVM (particle swarm optimization (PSO) and support vector machine techniques (SVM)). The evaluation indexes are constructed using data measured from 2018 to 2020 in six aspects: economy, science and technology, society, natural environment, resources, and ecological governance. 【Result】①The mean square error and the correlation coefficient of the wavelet-PSOSVM are 9.411 5×10-5 and 0.968, respectively. In contrast, the mean square error and the correlation coefficient of PSOSVM are 0.015 3 and 0.967, respectively, using the same training and test set. The wavelet process improves the predicting accuracy and convergence of the PSOSVM. ②The mean square error and the correlation coefficient of the wavelet-SVM are 20.836 and 0.748 respectively, while the mean square error and correlation coefficient of the SVM are 30.903 and 0.634, respectively, using the same training and test sets. The wavelet process also improvs the predicting accuracy and convergence of the SVM.【Conclusion】Using wavelet process in both PSOSVM and SVM algorithms significantly improves their prediction accuracy, convergence, and training time efficiency, while reducing their complexity. Its application for evaluating sustainable utilization of agricultural resources in the four provinces in northwestern China indicates that, in terms of sustainability, Xinjiang comes to the top, followed by Gansu, with Shaanxi and Ningxia ranked in third and fourth, respectively. |
| Key words: wavelet analysis; particle swarm optimization; support vector machine; resource sustainability utilization assessment |
