摘要
为解决传统土壤锰污染程度预测模型预测精度不足的问题,研究在融合小波神经网络与协同鸟群算法的基础上构建了深度复合神经网络的土壤锰污染程度预测模型。对研究提出的融合算法进行性能测试,结果显示研究提出的融合算法在误差小于10%及10%~20%的占比为75%~76%,该算法误差性能优于其他算法。对基于深度复合神经网络预测模型进行性能对比实验,结果显示其在四川和重庆2个数据集上的运算时间分别为26.6 s和24.5 s,较其他模型运算时间短,且其收敛速度更快。综合以上结果可以发现,研究提出融合算法及深度复合神经网络土壤锰污染程度预测模型在运算速度、运算精确度上优于对比算法与模型,具有实际应用价值。
To solve the problem of insufficient prediction accuracy in traditional soil heavy metal manganese(Mn)pollution prediction models,a deep composite neural network prediction model is proposed by integrating wavelet neural network and assisted bird swarm algorithm.Performance testing was conducted on the fusion algorithm proposed in the study.The results show that the proportion of errors between less than 10%and 10%-20%was 75%-76%.The error performance of this algorithm is superior to other algorithms.In addition,the study also conducted performance comparison experiments on deep composite neural network prediction models.The results showed that its computation time on the Sichuan and Chongqing datasets was 26.6 s and 24.5 s,respectively.Compared to other models,it has shorter computation time and faster convergence speed.Based on the above results,it can be found that the proposed fusion algorithm and deep composite neural network model for predicting the degree of heavy metal manganese(Mn)pollution are superior to the comparison algorithm and model in terms of computational speed and accuracy,which has practical application value.
作者
秦阳
李欣航
QIN Yang;LI Xinhang(Zoomlion Heavy Industry Co.,Ltd.,Changsha,Hunan 410013,China;Central South University of Forestry and Technology,Changsha,Hunan 410001,China)
出处
《中国锰业》
2023年第6期80-86,共7页
China Manganese Industry
关键词
小波神经网络
协同鸟群算法
深度复合神经网络
土壤锰污染
预测模型
Wavelet neural network
Collaborative bird swarm algorithm
Deep composite neural network model
Heavy metal pollution
Prediction model
