The precise measurement of Al, Mg, Ca, and Zn composition in copper slag is crucial for effective process control of copper pyrometallurgy. In this study, a remote laser-induced breakdown spectroscopy(LIBS) system was...The precise measurement of Al, Mg, Ca, and Zn composition in copper slag is crucial for effective process control of copper pyrometallurgy. In this study, a remote laser-induced breakdown spectroscopy(LIBS) system was utilized for the spectral analysis of copper slag samples at a distance of 2.5 m. The composition of copper slag was then analyzed using both the calibration curve(CC) method and the partial least squares regression(PLSR) analysis method based on the characteristic spectral intensity ratio. The performance of the two analysis methods was gauged through the determination coefficient(R^(2)), average relative error(ARE), root mean square error of calibration(RMSEC), and root mean square error of prediction(RMSEP). The results demonstrate that the PLSR method significantly improved both R^(2) for the calibration and test sets while reducing ARE, RMSEC, and RMSEP by 50% compared to the CC method. The results suggest that the combination of LIBS and PLSR is a viable approach for effectively detecting the elemental concentration in copper slag and holds potential for online detection of the elemental composition of high-temperature molten copper slag.展开更多
针对污水处理过程出水总磷预测问题存在的强非线性、大时变等特征,提出了一种基于偏最小二乘回归自适应深度信念网络(partial least square regression adaptive deep belief network,PLSR-ADBN)的出水总磷预测方法。PLSR-ADBN是基于深...针对污水处理过程出水总磷预测问题存在的强非线性、大时变等特征,提出了一种基于偏最小二乘回归自适应深度信念网络(partial least square regression adaptive deep belief network,PLSR-ADBN)的出水总磷预测方法。PLSR-ADBN是基于深度学习模型DBN的一种改进型建模方法。首先,将自适应学习率引入到DBN的无监督预训练(pre-training)阶段,来提高网络收敛速度。其次,利用PLSR方法取代传统DBN中基于梯度的逐层权值精调(fine-tuning)方法,来提高网络预测精度。同时,通过构造李雅普诺夫函数证明了PLSR-ADBN学习过程的收敛性。最后,将PLSR-ADBN用于实际污水处理过程出水总磷预测中。实验结果表明所提出的PLSR-ADBN收敛速度快且预测精度高,能够满足实际污水处理过程对出水总磷监测精度和运行效率的要求。展开更多
基金supported by funding for research activities of postdoctoral researchers in Anhui Provincespecial funds for developing Anhui Province’s industrial “three highs” and high-tech industries。
文摘The precise measurement of Al, Mg, Ca, and Zn composition in copper slag is crucial for effective process control of copper pyrometallurgy. In this study, a remote laser-induced breakdown spectroscopy(LIBS) system was utilized for the spectral analysis of copper slag samples at a distance of 2.5 m. The composition of copper slag was then analyzed using both the calibration curve(CC) method and the partial least squares regression(PLSR) analysis method based on the characteristic spectral intensity ratio. The performance of the two analysis methods was gauged through the determination coefficient(R^(2)), average relative error(ARE), root mean square error of calibration(RMSEC), and root mean square error of prediction(RMSEP). The results demonstrate that the PLSR method significantly improved both R^(2) for the calibration and test sets while reducing ARE, RMSEC, and RMSEP by 50% compared to the CC method. The results suggest that the combination of LIBS and PLSR is a viable approach for effectively detecting the elemental concentration in copper slag and holds potential for online detection of the elemental composition of high-temperature molten copper slag.
文摘针对污水处理过程出水总磷预测问题存在的强非线性、大时变等特征,提出了一种基于偏最小二乘回归自适应深度信念网络(partial least square regression adaptive deep belief network,PLSR-ADBN)的出水总磷预测方法。PLSR-ADBN是基于深度学习模型DBN的一种改进型建模方法。首先,将自适应学习率引入到DBN的无监督预训练(pre-training)阶段,来提高网络收敛速度。其次,利用PLSR方法取代传统DBN中基于梯度的逐层权值精调(fine-tuning)方法,来提高网络预测精度。同时,通过构造李雅普诺夫函数证明了PLSR-ADBN学习过程的收敛性。最后,将PLSR-ADBN用于实际污水处理过程出水总磷预测中。实验结果表明所提出的PLSR-ADBN收敛速度快且预测精度高,能够满足实际污水处理过程对出水总磷监测精度和运行效率的要求。
