Adaptive air-fuel ratio control of dual-injection engines under biofuel blends using extreme learning machine

Dual-injection engines, which allow real-time control and injection of two different fuels, are capable of varying the ratio of biofuel blends at different engine operating conditions for optimal engine performance. However, while many experiments have been carried out on these engines to demonstrate their advantages, very few studies have focused on the corresponding air-fuel ratio (AFR) control strategy. In order to achieve stable engine operation, it is essential to maintain transient AFR during the change of fuel blend ratio. Therefore, this study proposes an adaptive controller for AFR control of dual-injection engines. The proposed controller is designed based on a recently developed machine learning method called extreme learning machine, and its stability is verified with Lyapunov analysis. Simulations have been performed on an industry-level engine simulation software to verify the controller. Since dual-injection engines are not available in the market, a spark-ignition engine has been retrofitted for dual-injection operation so that the proposed controller can be implemented and evaluated experimentally. Both simulation and experiment results show that the proposed controller can effectively regulate the AFR to desired level. The results also show that the proposed controller outperforms the engine built-in AFR controller, indicating its significance for dual-injection engines.