1. Construction of digital twin model of engine in-cylinder combustion based on data-driven.
- Author
-
Hu, Deng, Wang, Hechun, Yang, Chuanlei, Wang, Binbin, Duan, Baoyin, Wang, Yinyan, and Li, Hucai
- Subjects
- *
DIGITAL twins , *COMBUSTION , *OPTIMIZATION algorithms , *ENGINE maintenance & repair , *PARAMETER identification - Abstract
Optimizing the combustion process by predicting combustion parameters during prolonged engine operation is crucial for engine maintenance. This study presents a zero-dimensional (0-D) prediction model that integrates the advantages of model-driven and data-driven approaches. Initially, the snake optimization algorithm (SO) is employed to address the challenges related to low parameter fitting accuracy and multiple solutions in calculating Wiebe parameters. Subsequently, a convolutional neural network-bidirectional long short-term memory neural network (CNN–Bi-LSTM) is devised to establish a nonlinear correlation between operating parameters and Wiebe parameters. The structural parameters of CNN–Bi-LSTM are then optimized using the SO algorithm (SO–CNN–Bi-LSTM). Ultimately, a 0-D prediction combustion model is formulated by amalgamating the Wiebe function with the neural network, enabling real-time prediction of combustion results and generalization analysis of prediction performance under non-calibrated conditions. The findings demonstrate that the combustion model exhibits heightened accuracy, thereby establishing a robust technical foundation for the development of a digital twin in the engine combustion process. • The SO algorithm for calculating Wiebe parameters was applied. • A parameter identification model based on SO–CNN–Bi-LSTM was proposed. • A 0-D prediction model was constructed by combining Wiebe function with neural network. • Based on the prediction model, the combustion process was simplified and reconstructed. [ABSTRACT FROM AUTHOR]
- Published
- 2024
- Full Text
- View/download PDF