基于台架采集数据,采用外部输入非线性自回归(nonlinear autoregressive model with exogenous input,NARX)神经网络建立了具备瞬态特性的柴油机排气温度计算模型作为虚拟传感器,并采用并发式训练方法对模型进行训练。将结果与前馈神经...基于台架采集数据,采用外部输入非线性自回归(nonlinear autoregressive model with exogenous input,NARX)神经网络建立了具备瞬态特性的柴油机排气温度计算模型作为虚拟传感器,并采用并发式训练方法对模型进行训练。将结果与前馈神经网络、长短期记忆网络(long short term memory,LSTM)神经网络及量产发动机的排温传感器采集结果进行对比。经验证,稳态工况下,两种神经网络均能达到较高精度;欧洲瞬态循环(European transient cycle,ETC)工况下,NARX神经网络计算温度的最大偏差为6.6℃,量产发动机排温传感器测得温度最大偏差为45.9℃。NARX神经网络所需的计算时间约为现有电控单元排温模型的2.5倍。展开更多
Groundwater is important for managing the water supply in agricultural countries like Bangladesh. Therefore, the ability to predict the changes of groundwater level is necessary for jointly planning the uses of ground...Groundwater is important for managing the water supply in agricultural countries like Bangladesh. Therefore, the ability to predict the changes of groundwater level is necessary for jointly planning the uses of groundwater resources. In this study, a new nonlinear autoregressive with exogenous inputs(NARX) network has been applied to simulate monthly groundwater levels in a well of Sylhet Sadar at a local scale. The Levenberg-Marquardt(LM) and Bayesian Regularization(BR) algorithms were used to train the NARX network, and the results were compared to determine the best architecture for predicting monthly groundwater levels over time. The comparison between LM and BR showed that NARX-BR has advantages over predicting monthly levels based on the Mean Squared Error(MSE), coefficient of determination(R^2), and Nash-Sutcliffe coefficient of efficiency(NSE). The results show that BR is the most accurate method for predicting groundwater levels with an error of ± 0.35 m. This method is applied to the management of irrigation water source, which provides important information for the prediction of local groundwater fluctuation at local level during a short period.展开更多
文摘基于台架采集数据,采用外部输入非线性自回归(nonlinear autoregressive model with exogenous input,NARX)神经网络建立了具备瞬态特性的柴油机排气温度计算模型作为虚拟传感器,并采用并发式训练方法对模型进行训练。将结果与前馈神经网络、长短期记忆网络(long short term memory,LSTM)神经网络及量产发动机的排温传感器采集结果进行对比。经验证,稳态工况下,两种神经网络均能达到较高精度;欧洲瞬态循环(European transient cycle,ETC)工况下,NARX神经网络计算温度的最大偏差为6.6℃,量产发动机排温传感器测得温度最大偏差为45.9℃。NARX神经网络所需的计算时间约为现有电控单元排温模型的2.5倍。
文摘Groundwater is important for managing the water supply in agricultural countries like Bangladesh. Therefore, the ability to predict the changes of groundwater level is necessary for jointly planning the uses of groundwater resources. In this study, a new nonlinear autoregressive with exogenous inputs(NARX) network has been applied to simulate monthly groundwater levels in a well of Sylhet Sadar at a local scale. The Levenberg-Marquardt(LM) and Bayesian Regularization(BR) algorithms were used to train the NARX network, and the results were compared to determine the best architecture for predicting monthly groundwater levels over time. The comparison between LM and BR showed that NARX-BR has advantages over predicting monthly levels based on the Mean Squared Error(MSE), coefficient of determination(R^2), and Nash-Sutcliffe coefficient of efficiency(NSE). The results show that BR is the most accurate method for predicting groundwater levels with an error of ± 0.35 m. This method is applied to the management of irrigation water source, which provides important information for the prediction of local groundwater fluctuation at local level during a short period.
