Linear Parameter Varying 모델링 기법을 이용한 디젤엔진 과급 압력 제어에 관한 연구

Abstract

Modeling and control of diesel engines have grown significantly in recent years because modern diesel engines are becoming more and more complex systems in order to meet emission standards and fuel consumption and power demands. To reduce NOx emissions, exhaust gas recirculation (EGR) systems have been added. Common rail injection systems have been adopted to reduce exhaust emissions, to improve fuel consumption and to reduce noise. Turbochargers have been added to increase engine torque. As these new technologies are incorporated into the modern diesel engines, they provide a great deal of freedom for controlling the behavior of the diesel engines. The main purpose of diesel engine control is to provide the required engine torque with minimal fuel consumption under the constraint of satisfying the stringent emission standards. This requires optimal coordination of the fuel injection system, the turbocharger system, and the EGR system under steady state and transient engine operations. Meanwhile, the current diesel engine controllers rely heavily on lookup tables for which only the minimal on-line calculation is required. In order to cover a wide range of operating conditions, many more lookup tables are needed. For example, besides the static engine maps, tables of compensation for transient operation have also been incorporated into the controller. Therefore, the engine controller development process requires extensive table calibrations, which is a very time-consuming and costly process. As the calculating speed of microcontrollers gets faster, it becomes desirable to consider the possibility of incorporating more complex process models into the engine controller. As a result, the study of engine modeling and application of advanced control theory has grown significantly in recent years. In this research, the advanced LPV techniques of identification and control are investigated. The applications focus on air path system modeling as well as control issues. In diesel engines, there are many subsystems that can be approximated by LPV models, such as the dynamics of boost pressure, exhaust manifold pressure, and EGR mass flow. LPV techniques use the advantage of the on-line measurable scheduling variables to achieve higher model precision and control performance. The applications of LPV techniques for diesel engines presented in this dissertation are successful. LPV techniques have many merits over LTI methodology. With the help of LPV modeling and control strategies, the model precision and control performance for diesel engines can be significantly improved Based on the simplified LPV boost pressure model, a robust H∞ controller is synthesized in this research. The state space representation is derived to follow the design procedure of a H∞ controller, and weighting functions for improving control performance are added into the closed loop control system. These methods provide some useful results for improving the control performance and simplify the calibration procedure for real applications. The experimental results are very promising and indicate that the controller synthesis with the LPV model and a H∞ controller is suitable for this control problem. |최근 내연기관의 효율이 과거 어느 때보다 중시됨에 따라 디젤엔진에 대한 관심이 지속적으로 증가하고 있다. 디젤엔진은 그 자체적인 특성 상 열효율이 높고 저속토크가 우수하다는 장점을 가지지만, NOx와 PM과 같은 유해 배출물이 많이 발생할 뿐만 아니라 진동 및 소음이 가솔린 엔진에 비하여 많이 발생하기 때문에 승용차량 적용에 장애가 존재한다. 그러나, 세계적인 추세인 온실가스 저감에 발맞추어 CO2 배출량이 상대적으로 적은 디젤엔진의 효용성이 점차 증가하고 있으며, 이에 따라 디젤엔진을 장착한 승용차량의 수요 또한 꾸준히 증가하고 있다. 최근의 디젤엔진은 성능, 연비, 그리고 배기규제를 만족시키기 위해서 점점 더 복잡한 형태를 갖추어가고 있다. 안정적인 연소와 배기가스 저감을 위한 피에조 타입 커먼레일 직분사 시스템과 연소압력 제어 시스템을 비롯해서, PM 필터, 디젤 산화 촉매, LNT, SCR 등과 같은 배기가스 후처리 장치도 광범위하게 연구되고 있다. 뿐만 아니라, 디젤엔진의 성능, 연비를 향상시킴과 동시에 배기가스를 저감 시킬 수 있는 공기흡입 시스템에 대한 연구 역시 중요한 이슈로 자리잡고 있다. 특히 공기흡입 시스템에는 가변 형상 터보차져, 스월 밸브, 스로틀 밸브, 고압 EGR, 저압 EGR 시스템 등과 같이 다양한 신기술이 개발되고 있다. 이러한 신기술은 디젤엔진에 적용되어 획기적인 성능 향상을 가져왔지만, 그에 비례하여 제어 기술의 필요성이 동시에 높아지고 있다. 기존의 엔진 제어기는 그 성능의 한계와 신뢰성으로 인하여 참조표 기반의 제어 방식을 선호한다. 이 방법은 정상상태 및 과도상태에서 사용 가능한 참조표 및 보정항을 수많은 엔진실험을 통하여 얻어야 한다는 특징을 가진다. 특히 점차적으로 복잡해지고 있는 디젤엔진 시스템을 고려한다면, 참조표를 작성하기 위한 실험적인 노력은 기하급수적으로 증가할 수 밖에 없다. 이는 결국 엔진 개발 기간과 비용의 증가로 귀결된다. 반면, 모델기반 제어는 대상 시스템의 수학적 모델을 이용하여 그 제어기를 설계하는 방법이다. 모델기반 제어를 이용하면 복잡한 시스템의 내부적인 연관성을 직접적으로 고려할 수 있기 때문에 참조표 방식의 제어 방법보다 유기적인 제어 알고리즘의 설계가 가능하다. 또한, 참조표가 필요한 부분을 수학적 모델에서 미리 고려할 수 있으므로 기존 제어 알고리즘의 참조표를 대체하는 효과를 얻을 수 있고, 제어기 개발에 필요한 인력, 시간, 비용을 저감할 수 있다. 뿐만 아니라 만들어진 모델의 재사용성이 뛰어나고 시스템 변화에 적극적으로 대처가 가능하기 때문에 다양한 분야에서 제어기 개발의 중요한 요소기술로서 전세계적으로 많은 연구가 수행되고 있다. 이 연구에서는 수학적 모델을 기반으로 디젤엔진의 과급 압력을 제어하였다. 제어기 설계를 위한 과급 압력 모델은 Linear Parameter Varying (LPV) 기법을 이용하여 모델링 하였다. 제안된 과급 압력 모델은 물리적인 법칙을 기본으로 하며, 그 구조가 간단하면서도 정확도를 보장한다. 또한 제안된 모델을 이용하여 과급 압력 제어기를 설계하였으며, 엔진 실험을 통해서 그 성능을 검증하였다.; Modeling and control of diesel engines have grown significantly in recent years because modern diesel engines are becoming more and more complex systems in order to meet emission standards and fuel consumption and power demands. To reduce NOx emissions, exhaust gas recirculation (EGR) systems have been added. Common rail injection systems have been adopted to reduce exhaust emissions, to improve fuel consumption and to reduce noise. Turbochargers have been added to increase engine torque. As these new technologies are incorporated into the modern diesel engines, they provide a great deal of freedom for controlling the behavior of the diesel engines. The main purpose of diesel engine control is to provide the required engine torque with minimal fuel consumption under the constraint of satisfying the stringent emission standards. This requires optimal coordination of the fuel injection system, the turbocharger system, and the EGR system under steady state and transient engine operations. Meanwhile, the current diesel engine controllers rely heavily on lookup tables for which only the minimal on-line calculation is required. In order to cover a wide range of operating conditions, many more lookup tables are needed. For example, besides the static engine maps, tables of compensation for transient operation have also been incorporated into the controller. Therefore, the engine controller development process requires extensive table calibrations, which is a very time-consuming and costly process. As the calculating speed of microcontrollers gets faster, it becomes desirable to consider the possibility of incorporating more complex process models into the engine controller. As a result, the study of engine modeling and application of advanced control theory has grown significantly in recent years. In this research, the advanced LPV techniques of identification and control are investigated. The applications focus on air path system modeling as well as control issues. In diesel engines, there are many subsystems that can be approximated by LPV models, such as the dynamics of boost pressure, exhaust manifold pressure, and EGR mass flow. LPV techniques use the advantage of the on-line measurable scheduling variables to achieve higher model precision and control performance. The applications of LPV techniques for diesel engines presented in this dissertation are successful. LPV techniques have many merits over LTI methodology. With the help of LPV modeling and control strategies, the model precision and control performance for diesel engines can be significantly improved Based on the simplified LPV boost pressure model, a robust H∞ controller is synthesized in this research. The state space representation is derived to follow the design procedure of a H∞ controller, and weighting functions for improving control performance are added into the closed loop control system. These methods provide some useful results for improving the control performance and simplify the calibration procedure for real applications. The experimental results are very promising and indicate that the controller synthesis with the LPV model and a H∞ controller is suitable for this control problem.