三江平原土壤质量变化评价的模糊神经网络模型构建及应用

针对土壤质量变化评价指标值的选取具有模糊性的特点，将BP网络和模糊数学相结合建立了土壤质量评价的模糊BP网络模型。选取三江平原具有典型性土壤质量变化趋势的试验点——富锦市永富乡土壤进行取样分析，选择与土壤质量相关的指标因子和相应的等级值建立评价模型。结果表明，该模型具有很强的学习能力，对参数的改变也有较强的鲁棒性，而且各个评价指标的权重都是由网络通过学习自动生成的，避免了主观赋权的人为干扰，因此评价结果更具有客观性和准确性，同时也为该领域研究提供了新的思路。

Construction and Application of Fuzzy Neural Networks Model in Evaluation on the Soil Quality Changes of Sanjiang Plain

Choosing evaluation targets of soil quality changing has fussiness,so fussy BP network model is created through the combination of BP network and fussy mathematics.Experimental area is in Yongfu town,Fujin city,in Sanjiang plain where the trend of soil quality changing is typical.Evaluation model is created through choosing the factor of soil quality changing and corresponding class value.The results show learning capability of the model is well,and the model manifests well robustness with parameter changing.The weight of evaluation target generates automatically through network itself studying,avoids the human interferences of weighting subjectively.So the evaluation result is objective and accurate.The model offered new idea for the research in this domain.