针对传统滑坡位移预测模型存在对历史数据遗忘的问题,提出了一种基于长短时记忆(long short time memory,LSTM)网络的滑坡位移动态预测模型。首先,将滑坡累计位移分解为趋势项位移与波动项位移,利用多项式拟合预测趋势项位移;然后,通过...针对传统滑坡位移预测模型存在对历史数据遗忘的问题,提出了一种基于长短时记忆(long short time memory,LSTM)网络的滑坡位移动态预测模型。首先,将滑坡累计位移分解为趋势项位移与波动项位移,利用多项式拟合预测趋势项位移;然后,通过灰色关联度筛选外界诱发因子并运用LSTM模型预测波动项位移;最后,叠加周期项位移与波动项位移,得到累计位移。以新滩滑坡为例,并与(recurrent neural network,RNN)模型以及传统静态神经网络模型BP、ELM进行对比分析,采用平均百分比误差(MAPE),均方根误差(RMSE),拟合优度(R 2)分别对其进行评价。应用结果表明:相比于传统静态模型,LSTM与RNN均适用于滑坡位移动态预测;对比结果显示,LSTM模型具有较好的预测精度,MAPE与RMSE分别为1.026%、0.327 mm,拟合优度R 2为0.978。展开更多
ith an improved EHMO method, three modes (flat-lying, vertical , and in-clined insertion) for adsorption and activation of dioxygen on (100) , (110) and(111) surfaces of Na_2O and of K,O have been exaniinecl, and the ...ith an improved EHMO method, three modes (flat-lying, vertical , and in-clined insertion) for adsorption and activation of dioxygen on (100) , (110) and(111) surfaces of Na_2O and of K,O have been exaniinecl, and the interaction ofthese dioxygen adspecies with CH_4 and with CH_3 · (radical) from gas pliase liasbeen investigated. The results indicate that both oxides tend to forni less cliargedadspecies of dioxygen, with the flat-lying adsorption on (110) surface most favor-able energetically. All these cliemisorbed dioxygen species are capable of interactingeffectively with CH_4 and Abstract ing one hydrogen atom from the CH_4 molecule andtheir tendencies to reassociate with CH_3 · , which would easily lead to deep oxida-tion of the fragments of hydrocarbons, are enhanced with inereasing negativecliarges on them. In comparison with Na_2O, K_2O has a little stronger tendency tostabilize less charged dioxygen adspecies; this has a close relation with the knownexperimental fact that K￣+ showed better promoting effect than Na￣+ in improvingC_2-selectivity in methane oxidative coupling (MOC ).展开更多
文摘针对传统滑坡位移预测模型存在对历史数据遗忘的问题,提出了一种基于长短时记忆(long short time memory,LSTM)网络的滑坡位移动态预测模型。首先,将滑坡累计位移分解为趋势项位移与波动项位移,利用多项式拟合预测趋势项位移;然后,通过灰色关联度筛选外界诱发因子并运用LSTM模型预测波动项位移;最后,叠加周期项位移与波动项位移,得到累计位移。以新滩滑坡为例,并与(recurrent neural network,RNN)模型以及传统静态神经网络模型BP、ELM进行对比分析,采用平均百分比误差(MAPE),均方根误差(RMSE),拟合优度(R 2)分别对其进行评价。应用结果表明:相比于传统静态模型,LSTM与RNN均适用于滑坡位移动态预测;对比结果显示,LSTM模型具有较好的预测精度,MAPE与RMSE分别为1.026%、0.327 mm,拟合优度R 2为0.978。
文摘ith an improved EHMO method, three modes (flat-lying, vertical , and in-clined insertion) for adsorption and activation of dioxygen on (100) , (110) and(111) surfaces of Na_2O and of K,O have been exaniinecl, and the interaction ofthese dioxygen adspecies with CH_4 and with CH_3 · (radical) from gas pliase liasbeen investigated. The results indicate that both oxides tend to forni less cliargedadspecies of dioxygen, with the flat-lying adsorption on (110) surface most favor-able energetically. All these cliemisorbed dioxygen species are capable of interactingeffectively with CH_4 and Abstract ing one hydrogen atom from the CH_4 molecule andtheir tendencies to reassociate with CH_3 · , which would easily lead to deep oxida-tion of the fragments of hydrocarbons, are enhanced with inereasing negativecliarges on them. In comparison with Na_2O, K_2O has a little stronger tendency tostabilize less charged dioxygen adspecies; this has a close relation with the knownexperimental fact that K￣+ showed better promoting effect than Na￣+ in improvingC_2-selectivity in methane oxidative coupling (MOC ).
