Model-free adaptive control scheme for EGR/VNT control of a diesel engine using the simultaneous perturbation stochastic approximation. (January 2017)

• Record Type: Journal Article
• Authors: Ishizuka, Shinichi; Kajiwara, Itsuro; Sato, Junichi; Hanamura, Yoshifumi; Hanawa, Satoshi
• Is Part Of: Transactions of the Institute of Measurement and Control. Volume 39:Number 1(2017)
• Page Start: 114
• Page End: 128
• Publication Date: 2017-01
• Subjects: Adaptive control -- diesel engine -- EGR/VNT -- model-free control -- PID control -- simultaneous perturbation stochastic approximation; Automatic control -- Periodicals; Measuring instruments -- Periodicals; Commande automatique -- Périodiques; Mesure -- Instruments -- Périodiques; 681.2
• DOI: 10.1177/0142331215602327
• Languages: English
• ISSNs: 0142-3312

Abstract

To meet the more stringent environmental requirements of automobile exhaust gas emissions, diesel engines have recently received increased attention due to their high heat efficiency. To lower fuel consumption and reduce exhaust gas simultaneously, fuel combustion must be more precisely controlled. For example, the oxygen concentration, which affects emissions, is controlled by exhaust gas recirculation (EGR) and variable nozzle turbo (VNT). However, realizing a controlled design is difficult due to system non-linearity and strong interference between EGR and VNT. Recently, various design methods have employed the so-called model-based control design, but this design approach is difficult to use when the controlled object is complex. Currently, mass production uses gain scheduling of map-based on proportional–integral–derivative (PID) control, in which each gain is tuned at various operational points. However, map calibration has many drawbacks, including time-consuming tuning, difficulty tuning during transient operations and problems adapting to the individual variations in the engine characteristics. This study proposes a construction method for a model-free adaptive PID controller using the simultaneous perturbation stochastic approximation (SPSA) and its performance is confirmed in an engine bench test.