基于数字孪生的船用柴油机燃烧模型的构建与验证
Construction and verification of combustion model of marine diesel engine based on digital twin
船用柴油机燃烧模型的构建是对处于长时间运行状态的发动机的性能优化和健康管理的关键手段。本文基于Wiebe燃烧模型结合深度学习神经网络提出了基于混合驱动的发动机零维预测模型构建方案，用于实现发动机的同步仿真。首先对通过试验获得的缸压曲线进行Wiebe参数求解，进而利用长短时记忆神经网络（LSTM）建立运行参数与Wiebe参数的辨识模型，然后将Wiebe方程与深度学习神经网络相结合构建出零维预测燃烧模型，并对预测性能进行非校核工况的泛化性分析。基于数字孪生的发动机燃烧模型是实现发动机燃烧过程在线预测的一种有效方法，同时为未来的在线优化提供新的理论依据。
The construction of marine diesel engine combustion model is a key means for performance optimization and health management of the engine in a long-term running state. Based on the Wiebe combustion model combined with the deep learning neural network, this paper proposes a zero-dimensional engine prediction model construction scheme based on hybrid drive, which is used to realize the synchronous simulation of the engine. First, solve the Wiebe parameters for the cylinder pressure curve obtained through the test, and then use the long-short-term memory neural network (LSTM) to establish the identification model of operating parameters and Wiebe parameters, and then combine the Wiebe equation with the deep learning neural network to construct a zero-dimensional predictive combustion model, and conduct a generalization analysis of the predictive performance under non-calibrated working conditions. The engine combustion model based on the digital twin is an effective method to realize the online prediction of the engine combustion process, and at the same time provide a new theoretical basis for future online optimization.
胡登;王贺春;杨传雷;王彬彬;段宝印;王银燕;
哈尔滨工程大学动力与能源工程学院,哈尔滨150001;哈尔滨工程大学动力与能源工程学院,哈尔滨150001;哈尔滨工程大学动力与能源工程学院,哈尔滨150001;哈尔滨工程大学动力与能源工程学院,哈尔滨150001;哈尔滨工程大学动力与能源工程学院,哈尔滨150001;哈尔滨工程大学动力与能源工程学院,哈尔滨150001;
diesel engine digital twin LSTM neural network burning model
1-7 / 7
评论