中文摘要：

在小型浆体流动试验系统上采用4根不同管径的直管考察水焦浆的阻力特性。水焦浆在管内流动存在壁面滑移效应,具有滑移减阻现象,压降预测需要进行壁面滑移修正。利用粒子群优化算法（particle swarm optimization,PSO）对BP神经网络进行改进,建立考虑5因子影响因素后的水焦浆管道输送压降PSO-BP神经网络预测模型;采用神经网络预测模型对水焦浆在管道输送中的压降进行了预测,并将预测值与试验值进行比较。结果表明：粒子群优化算法改进的神经网络模型可以有效预测水焦浆在管道输送过程中的压降,预测值与试验值之间误差较小,平均绝相对误差不超过10%。

英文摘要：

Experimental researches were carried out on a small-scale slurry transportation system to investigate resistance properties of coke water slurry flowing in pipes with four different diameters.There exists wall slip behavior when coke water slurry flows in the pipes,which could cause drag reduction.Therefore,it is necessary to correct wall slip behavior to predict pressure drop of coke water slurry.An artificial neural networks improved by PSO（particle swarm optimization） with five parameters in input layer was constructed to predict the pressure drop of coke water slurry flowing in the pipeline.Then pressure drop was predicted by artificial neural network and results were compared with the experimental value.The results show that PSO-BP artificial neural network has a good ability in predicting the pressure drop of coke water slurry flowing in the pipe.The error between the predicted value and experimental value is small and the largest error is no more than 10%.