Control-oriented modelling of a Wankel rotary engine: a synthesis approach of state space and neural networks

Anthony Siming Chen, Giovanni Vorraro, Matthew Turner, Reza Islam, Guido Herrmann, Stuart Burgess, Chris Brace, Jamie Turner, Nathan Bailey

Research output: Contribution to conferencePaper

Abstract

The use of Wankel rotary engines as a range extender has been
recognised as an appealing method to enhance the performance of
Hybrid Electric Vehicles (HEV). They are effective alternatives to
conventional reciprocating piston engines due to their considerable
merits such as lightness, compactness, and higher power-to-weight
ratio. However, further improvements on Wankel engines in terms of
fuel economy and emissions are still needed. The objective of this
work is to investigate the engine modelling methodology that is
particularly suitable for the theoretical studies on Wankel engine
dynamics and new control development.
In this paper, control-oriented models are developed for a 225CS
Wankel rotary engine produced by Advanced Innovative Engineering
(AIE UK) Ltd. Through a synthesis approach that involves State Space
(SS) principles and the artificial Neural Networks (NN), the Wankel
engine models are derived by leveraging both first-principle
knowledge and engine test data. We first re-investigate the classical
physics-based Mean Value Engine Model (MVEM). It consists of
differential equations mixed with empirical static maps, which are
inherently nonlinear and coupled. Therefore, we derive a SS
formulation which introduces a compact control-oriented structure
with low computational demand. It avoids the cumbersome structure
of the MVEM and can further facilitate the advanced modern control
design. On the other hand, via black-box system identification
techniques, we compare the different NN architectures that are suitable
for engine modelling using time-series test data: 1) the Multi-Layer
Perceptron (MLP) feedforward network; 2) the Elman recurrent
network; 3) the Nonlinear AutoRegressive with eXogenous inputs
(NARX) recurrent network. The NN models overall tend to achieve
higher accuracy than the MVEM and the SS model and do not require
a priori knowledge of the underlying physics of the engine.
Original languageEnglish
Number of pages14
Publication statusAcceptance date - 27 Jan 2020
EventSAE World Congress 2020 - COBO Centre, Detroit, USA United States
Duration: 21 Apr 202023 Apr 2020
https://www.sae.org/attend/wcx

Conference

ConferenceSAE World Congress 2020
Abbreviated titleSAE WCX
CountryUSA United States
CityDetroit
Period21/04/2023/04/20
Internet address

Cite this

Chen, A. S., Vorraro, G., Turner, M., Islam, R., Herrmann, G., Burgess, S., Brace, C., Turner, J., & Bailey, N. (Accepted/In press). Control-oriented modelling of a Wankel rotary engine: a synthesis approach of state space and neural networks. Paper presented at SAE World Congress 2020, Detroit, USA United States.