Detection of transient low-temperature combustion characteristics by ion current – The missing link for homogeneous charge compression ignition control?
Detection of transient low-temperature combustion characteristics by ion current – The missing link for homogeneous charge compression ignition control?
复制标题
DOI：
10.1016/j.apenergy.2020.116299
复制
发表时间：
2020-12
期刊：
影响因子：
11.2
通讯作者：
Jinqiu Wang;Julian Bedei;J. Deng;J. Andert;Denghao Zhu;Liguang Li
中科院分区：
文献类型：
--
作者：
Jinqiu Wang;Julian Bedei;J. Deng;J. Andert;Denghao Zhu;Liguang Li
均质充量压燃具有高效、低排放等优点，是一种很有发展前途的低温燃烧方式，但由于废气再循环引起的强循环耦合效应，可能导致燃烧稳定性降低。在这项研究中，一个新的概念，提取更全面的燃烧信息，在均质充量压燃提出了通过集成的离子电流和缸内压力传感。为了分析燃烧参数的相关性及其与离子电流参数的关系，进行了稳态测量。实现动态测量以形成用于人工神经网络训练的综合数据库。为了研究离子电流提供压力迹线之外的额外信息的假设，训练基于实验数据的黑盒模型。结果表明，单纯使用操纵变量训练的基准模型性能最差，而同时包含缸内压力和离子电流导出参数的模型具有最好的预测性，与基于缸内压力的模型相比，预测燃烧相位的总均方根误差降低了2.5%。它表明，一个显着的改进模型的质量可以实现通过离子电流和缸内压力传感的组合，这表明，离子电流信号包含的信息，超越了压力轨迹的单独分析。通过补充缸内压力，使用离子电流作为低温燃烧的“化学传感器”，因此对于均质充量压燃燃烧的稳定控制显得非常有希望。
Homogeneous charge compression ignition is a promising low-temperature combustion mode because of its high efficiency and low emissions; however, its strong cycle-to-cycle coupling effect, which caused by the recirculation of exhaust gases, may entail problems with low combustion stability. In this study, a new concept that extracts more comprehensive combustion information in homogeneous charge compression ignition is proposed through the integration of ion current and in-cylinder pressure sensing. To analyze the correlations of combustion parameters and their relationships with the ion current parameters, steady-state measurements were conducted. Dynamic measurements were implemented to form a comprehensive database for artificial neural network training. To investigate the hypothesis that the ion current gives additional information beyond the pressure trace, black-box models based on experimental data are trained. The results show that the baseline model trained purely with the manipulated variables has the worst performance, while the model including both in-cylinder pressure and ion current derived parameters has the best predictability, with the overall root-mean-square error reduced by 2.5% in predicting combustion phasing, compared with in-cylinder pressure based model. It demonstrates that a significant improvement in model quality can be achieved by the combination of ion current and in-cylinder pressure sensing, which indicates that the ion current signal contains information that goes beyond a sole analysis of the pressure trace. By complementing the in-cylinder pressure, the use of the ion current as a “chemical sensor” for low-temperature combustion thus appears very promising for the stable control of homogeneous charge compression ignition combustion.