采用BP神经网络对气力输送中固体流量测量的预测研究

Investigation on Prediction of Solid Flow-rate in Gas-solid Two Phase by BP Neural Networks

采用传热法测量稀相气固两相流中固相流量时,固相流量与其影响因素之间存在着复杂的非线性关系,给实际工业应用带来许多困难。本文利用人工神经网络优良的非线性映射能力,设计了一个三层BP神经网络,以实验数据为样本对网络进行训练后,预测气固两相流中的固相流量,预测结果和实验结果吻合较好,为稀相气力输送中固相流量测量提供了一种简单、可靠的新方法。

A methodology is introduced to use neural networks for online measurement of the solid flow-rate in gas-solid two phase flow based on heat transfer. An electrically heated probe was put in a gas-solid two-phase flow. The flow mediums with different velocity of flow, densities and diameters of particles produced different results of heat transfer. For a certain velocity of conveyer air, the heating electric power and the superficial temperature of the probe could determine the solid flow rate. Experiments were made on a pilot gas-solid conveyer device. Prediction results prove that the method works effectively and reliably.