考虑壁面影响的矩形裂缝通道中的颗粒沉降

为了表征颗粒在矩形裂缝通道中沉降时壁面对沉降速度的影响,基于人工神经网络(ANN)提出壁面因子预测方法,提取影响颗粒沉降速度的7个参数(ρf,ρp,d,d/a,a/b,K和n)作为特征值,借助Machac的70组试验数据对模型进行训练和预测,并将人工神经网络预测结果与Miyamura公式和刘马林公式计算结果进行对比分析.结果表明:人工神经网络模型具有较高的精度,90%的预测结果误差小于7.5%.与Miyamura公式和刘马林公式相比,人工神经网络模型不但在处理平行板模型和矩形模型时有较高的工程精度,而且具有更广泛的适用范围,能够满足更加复杂的工程需要.

Research on particle settling in rectangular ducts under effect of walls

In order to characterize wall effect of rectangular ducts on particle settling, a prediction method of wall factor was presented using Artificial Neural Network(ANN). Extracted ρf, ρp, d, d/a, a/b, K and n as feature parameters, which substantially affect particle settling velocity.Culled 70 data of Machac′s paper to train and test the models, and compared predicted results of ANN with the calculated results of Miyamura correlation and Liu Malin correlation.The results show that, the accurate of ANN predicted results is great, the error is less than 7.5% for 90% calculated results. Comparing to Miyamura correlation and Liu Malin correlations, the ANN model has high engineering accuracy in parallel plate and rectangular ducts, and the application range of ANN has broader scope of application,and can meet complex engineering needs.