基于人工神经网络的油水旋流分离器分离性能预测

Prediction of the separation performance of hydrocyclones for oil-water separation by artificial neural network model

采用改进的BP神经网络模型模拟水力旋流器的油水分离过程.根据水力旋流器的实际运行条件,确定旋流器模型设计中的优化神经网络结构,将遗传算法用于优化三层BP神经网络的初始权重,采用PRP共轭梯度法优化BP算法.结果表明,采用人工神经网络模型预测油水分离水力旋流器的分离性能是切实可行的,它能成功地模拟旋流器的分离过程,进而实现旋流器操作控制的优化.

Oil-water separation in hydrocyclones was simulated by an improved back propagation （BP） neural network model. According to the actual hydrocyclones operating condition , an optimized structure of BP neural network was ascertained. Genetic algorithm was used to optimize the initial weight and the polak-ribière-polyak （PRP） conjugated method was used to optimize the algorithm of BP neural network. The results indicate that predicting the separation performance of hydrocyclones by means of artificial neural network is feasible. It can simulate the oil-water separation process efficiently. And the optimization of the operation of hydrocyclones can be realized by the simulation of BP neural network.