CAJ | 학술논문

NdF3-LiF-Nd2O3体系熔盐电导率是稀土熔盐电解的基础参数，由于高温环境使其在电解过程中的变化规律难以获得。研究针对试验结果所取得的样本进行训练，通过BP神经网络预测了NdF3-LiF-Nd2O3体系熔盐电导率，并分析了温度、LiF浓度和Nd2O3浓度对熔盐电导率的影响。研究结果表明，预测值处在1.8256~3.1197 S/cm之间，与实验值误差在3%左右，同时，预测值与实际值的变化趋势基本一致。研究表明BP神经网络对熔盐电导率的预测能够满足熔盐体系电导率研究的要求。
NdF3-LiF-Nd2O3체계용염전도솔시희토용염전해적기출삼수，유우고온배경사기재전해과정중적변화규률난이획득。연구침대시험결과소취득적양본진행훈련，통과BP신경망락예측료NdF3-LiF-Nd2O3체계용염전도솔，병분석료온도、LiF농도화Nd2O3농도대용염전도솔적영향。연구결과표명，예측치처재1.8256~3.1197 S/cm지간，여실험치오차재3%좌우，동시，예측치여실제치적변화추세기본일치。연구표명BP신경망락대용염전도솔적예측능구만족용염체계전도솔연구적요구。
The basic parameter of rare earth electrolys is the electrical conductivity of NdF3-LiF-Nd2O3 molten salts. However, there is difficult to obtain the changes of the process of electrolysis because of the high temperature environment. With the experimental results obtained in the research on the training samples, to predict the conductivity of NdF3-LiF-Nd2O3 molten salts by the BP neural network, and analyze the influence of temperature, LiF and Nd2O3 on the electrical conductivity of molten salts. The research results show that the predicted values in 1.825 6~3.119 7 S·cm-1, and the experimental value of error is about 3%, predictive value and the actual value of the change tendency is consistent. Researches show that prediction of conductivity of molten salts by BP neural network can meet the requirements of electrical conductivity of molten salts system.