船用柴油机实时缸压的混合建模研究
Research on hybrid modeling of real-time cylinder pressure in Marine diesel engines
实时缸压数据是实现柴油机循环控制的重要参数,直接通过压力传感器测量成本过高。模型估计法和间接测量法是获取缸内压力数据的两种常见方式,但这两种方法是在牺牲实时性的前提下保证其准确性。为了获取实时准确的缸内压力,提出基于数据驱动模型与动力学模型的船用柴油机混合建模方法。首先,通过历史运行缸压数据对韦伯燃烧参数进行标定并构建运行参数与韦伯燃烧参数非线性关系的柴油机数据驱动模型。接着,通过分析曲轴连杆系统的动力学特性,构建一个基于瞬时转速重构缸压的柴油机动力学模型。最后,利用加权平均法对数据驱动模型和动力学模型进行混合建模。结果表明,在不损失实时性的前提下,采用混合建模方法可获取更准确的缸压数据。
Real-time cylinder pressure data is an important parameter to realize cycle control of diesel engine, and it is too expensive to measure it directly by pressure sensor. Model estimation method and indirect measurement method are two common ways to obtain cylinder pressure data, but these two methods guarantee their accuracy at the expense of real-time. In order to obtain real-time and accurate in-cylinder pressure, a hybrid modeling method of Marine diesel engine based on data-driven model and dynamic model is proposed. Firstly, Wiebe combustion parameters are calibrated by historical operating cylinder pressure data, and a diesel data-driven model based on the nonlinear relationship between the operating parameters and Wiebe combustion parameters is constructed. Then, by analyzing the dynamic characteristics of the crankshaft connecting rod system, a diesel engine dynamic model based on instantaneous speed reconstruction of cylinder pressure is constructed. Finally, the weighted average method is used to build a hybrid model of data-driven model and dynamic model. The results show that the hybrid modeling method can obtain more accurate cylinder pressure data without loss of real-time performance.
李丽微;刘岱;
哈尔滨工程大学动力与能源工程学院,黑龙江 哈尔滨150001;哈尔滨工程大学动力与能源工程学院,黑龙江 哈尔滨150001;
TK421+.2
marine diesel engine cylinder pressure prediction data-driven model dynamic model hybrid
1-8 / 8
评论