This study was devoted to introducing a new method for a priori prediction of aspiration pressure buildup in closed coupled atomization (CCA) nozzles. There have been considerable controversies about increasing or decreasing the aspiration pressure for a reliable operation of CCA nozzles, mainly because of the complex nature of CCA process. Here for the first time, we applied an artificial neural network (ANN) based machine learning algorithm for the prediction of aspiration pressure in close-coupled HPGA nozzles. An analytical model equation was obtained based on the largest experimental dataset from the literature and proved to be useful for prediction of non-dimensionalized aspiration pressure with R² of 0.98. But, its prediction accuracy of absolute aspiration pressures was degraded with a decrease of R² score to 0.73 and an average prediction error of 17 %, mainly due to the limitation of literature data. Based on parametric study and a sensitivity test, protrusion length of CCA nozzles and Re number were found to be relatively significant as compared to the apex angles. Finally, we provided a comprehensive contour map to facilitate the conceptual design and operation of CCA nozzles to minimize the aspiration pressure.

这项研究致力于介绍一种新方法，用于对封闭耦合雾化（CCA）喷嘴中的吸气压力累积进行先验预测。为了CCA喷嘴的可靠操作，关于增加或减小抽吸压力存在很大争议，这主要是由于CCA工艺的复杂性。在这里，我们首次将基于人工神经网络（ANN）的机器学习算法用于预测紧密耦合HPGA喷嘴中的抽吸压力。基于文献中最大的实验数据集获得了一个解析模型方程，证明该模型方程对于预测R ²为0.98的无量纲抽吸压力很有用。但是，其绝对吸气压力的预测精度随R的降低而降低²得分为0.73，平均预测误差为17％，这主要是由于文献数据的限制。根据参数研究和敏感性测试，发现CCA喷嘴的突出长度和Re数与顶角相比相对较大。最后，我们提供了全面的轮廓图，以方便CCA喷嘴的概念设计和操作，以最大程度地减小抽吸压力。