Robust air-to-fuel ratio and boost pressure controller design for EGR and VGT systems using QFT method

Abstract

With newly developed fuel injection technologies such as high pressure common rail and injectors, the market of diesel engines has rapidly extended from commercial vehicles to passenger cars in the past decade. Along with the popularized passenger car diesel engines, the emission legislations have tightened in order to reduce particulate matter (PM) and NOx emissions which cause severe air pollution problems. In order to satisfy these stringent emission regulations and drivability requirements concurrently, the exhaust gas recirculation (EGR) and the variable geometry turbocharger (VGT) technologies play important roles in passenger car diesel engines. Therefore, improving the control performance of the EGR and VGT systems is recognized as a major challenge in the automotive industry. Unfortunately, the EGR and VGT systems have highly nonlinear characteristics. Furthermore, they are strongly interactive because each system forms a loop between the intake and exhaust manifolds. With successful handling of nonlinear characteristic of EGR and VGT, coordinated control is a delicate problem. This thesis proposed two kinds of design methodologies for the EGR and VGT systems of passenger car diesel engines. In this study, the air-to-fuel ratio of exhaust gas and the boost pressure of intake manifold are selected as performance variables. The first approach is the robust multi-input multi-output (MIMO) feedback design using quantitative feedback theory (QFT) method. In order to keep the accessibility of on-line calibration, the structure of the feedback controller was restricted to proportional-integral-derivative (PID) which is widely used in industry applications. Using quantitative feedback theory (QFT), two control loops for air-to-fuel ratio and boost pressure were independently designed with linearized model and structured parameter uncertainty. The prefilters and PID controllers of two control loops are designed for satisfying desired robust stability and tracking performance using the QFT design framework. Furthermore, problems originated from cross-coupled dynamics between the EGR and VGT systems are mitigated by using the static decoupler. The proposed controller design steps are applied to 15 significant engine operating points. Using gain-scheduling technique, the designed controllers and decouplers of entire operating points are implemented. The second design methodology is the non-linear model-based feedforward controller. Mean-value engine model of the EGR and VGT systems is proposed by using in-cylinder mass balance, mass conservation of intake and exhaust manifold, turbocharger efficiency and orifice valve models. Based on the proposed model, the feedforward algorithms of the EGR and VGT systems are derived by using static model inversion approach. Using the stationary measurements of 225 engine operating conditions, accurate parameterization results are obtained. It is observed that the proposed feedforward algorithms have 2% of mean modeling errors. In order to compensate the modeling errors of feedforward controller, PI-based feedback controller is used and the gains are tuned by engine experiments. The proposed two design methodologies are validated through the various operating conditions of engine experiments. From the experimental results of the robust MIMO feedback design methodology, the design results are satisfied the required tracking performance and robustness as expected. In the case of the second approach, the proposed feedforward controller successfully compensates the non-linear behavior of EGR and VGT systems during transient responses. | 커먼레일과 고압분사와 같은 새로운 연료분사 기술들이 적용 됨에 따라 최근 10년 간 유럽을 중심으로 승용디젤엔진 자동차 시장이 급격하게 성장하고 있다. 배기가스 배출 규제의 강화와 함께 질소산화물(NOx)과 입자상 물질(PM)의 배출은 승용디젤엔진의 중요한 문제로 인식되고 있으며, 운전성 확보와 배기가스 배출물 규제 만족이라는 목적들을 동시에 달성하기 위하여 가변용량 터보차저(Variable Geometry Turbocharger, VGT)와 배기가스 재순환(Exhaust Gas Recirculation, EGR) 과 같은 기술들이 적용되기 시작하여 현재는 널리 이용되고 있다. EGR과 VGT의 제어 성능이 엔진의 배기와 동력 성능에 큰 영향을 미치기 때문에 정교한 제어 알고리즘 설계가 매우 중요하다. 하지만, EGR과 VGT 시스템은 EGR valve와 VGT vane의 유효 단면적, 압력 비에 따른 질량 유량, 터보차저 효율 등과 같은 요인들로 인하여 강한 비선형적인 특징을 지니고 있다. 또한, 각각의 시스템들이 형성하고 있는 물리적인 경로들이 흡기와 배기 매니폴드들을 통해 연결되어 있어 서로 큰 영향을 미치는 다변수 시스템 (Multi-input multi-output system) 이다. 이와 같은 특징들로 인하여 EGR과 VGT 시스템의 제어알고리즘 설계에는 많은 어려움들이 존재한다. 이 연구에서는 승용디젤엔진의 EGR과 VGT 시스템을 위한 제어알고리즘 설계를 위한 두 가지 설계 기법에 대하여 제시하고 있다. 배기가스의 공연비와 흡기 매니폴드의 과급압력을 제어인자로 설정하였으며, EGR valve 와 VGT vane을 제어입력으로 사용하였다. 첫 번째 설계 기법은 quantitative feedback theory (QFT)를 이용한 강인 MIMO feedback 제어기 설계이다. 양산 제어기 적용을 가능케 하기 위하여 연산부하가 적고, on-line calibration 이 가능한 PID를 피드백 제어기의 구조로 고정하였다. 운전 영역에 따라 PID 제어기의 게인을 스케줄링 함으로써 발생하는 calibration, robustness 문제는 QFT 기반의 설계 기법을 이용하여 다양한 엔진 운전조건에서 강인 안정성과 성능을 확보하였다. 또한 EGR과 VGT 시스템의 강한 상호작용(interaction)은 정적 디커플러(static decoupler) 적용을 통해 보상하였다. 두 번째는 비선형 모델 기반 feedforward 알고리즘을 이용한 제어기 설계 기법이다. In-cylinder mass balance, mass conservation law, turbocharger efficiency 및 orifice valve 와 같은 모델들을 이용하여 EGR과 VGT 시스템의 비선형 평균 값 정적 모델들을 제시하였다. 제시한 모델들의 역변환을 통해 목표 공연비와 과급압력을 추종하기 위한 feedforward 알고리즘을 제시하였으며, 225 가지의 다양한 엔진 운전조건에서의 정상상태 실험데이터를 이용하여 평균 2%의 오차를 보이는 parameterization 결과를 얻을 수 있었다. PI 제어기를 추가함으로써, 정상상태 및 과도구간에서의 feedforward 제어기의 모델링 오차를 보상하도록 구현하였다. 이 연구에서 제시한 두 가지 제어기 설계 기법들은 각각 연구용으로 자체 개발한 엔진제어기 플랫폼을 이용하여 구현하였으며, 동력계 환경에서 다양한 조건의 다양한 엔진실험을 통해 그 성능들을 성공적으로 검증하였다.; With newly developed fuel injection technologies such as high pressure common rail and injectors, the market of diesel engines has rapidly extended from commercial vehicles to passenger cars in the past decade. Along with the popularized passenger car diesel engines, the emission legislations have tightened in order to reduce particulate matter (PM) and NOx emissions which cause severe air pollution problems. In order to satisfy these stringent emission regulations and drivability requirements concurrently, the exhaust gas recirculation (EGR) and the variable geometry turbocharger (VGT) technologies play important roles in passenger car diesel engines. Therefore, improving the control performance of the EGR and VGT systems is recognized as a major challenge in the automotive industry. Unfortunately, the EGR and VGT systems have highly nonlinear characteristics. Furthermore, they are strongly interactive because each system forms a loop between the intake and exhaust manifolds. With successful handling of nonlinear characteristic of EGR and VGT, coordinated control is a delicate problem. This thesis proposed two kinds of design methodologies for the EGR and VGT systems of passenger car diesel engines. In this study, the air-to-fuel ratio of exhaust gas and the boost pressure of intake manifold are selected as performance variables. The first approach is the robust multi-input multi-output (MIMO) feedback design using quantitative feedback theory (QFT) method. In order to keep the accessibility of on-line calibration, the structure of the feedback controller was restricted to proportional-integral-derivative (PID) which is widely used in industry applications. Using quantitative feedback theory (QFT), two control loops for air-to-fuel ratio and boost pressure were independently designed with linearized model and structured parameter uncertainty. The prefilters and PID controllers of two control loops are designed for satisfying desired robust stability and tracking performance using the QFT design framework. Furthermore, problems originated from cross-coupled dynamics between the EGR and VGT systems are mitigated by using the static decoupler. The proposed controller design steps are applied to 15 significant engine operating points. Using gain-scheduling technique, the designed controllers and decouplers of entire operating points are implemented. The second design methodology is the non-linear model-based feedforward controller. Mean-value engine model of the EGR and VGT systems is proposed by using in-cylinder mass balance, mass conservation of intake and exhaust manifold, turbocharger efficiency and orifice valve models. Based on the proposed model, the feedforward algorithms of the EGR and VGT systems are derived by using static model inversion approach. Using the stationary measurements of 225 engine operating conditions, accurate parameterization results are obtained. It is observed that the proposed feedforward algorithms have 2% of mean modeling errors. In order to compensate the modeling errors of feedforward controller, PI-based feedback controller is used and the gains are tuned by engine experiments. The proposed two design methodologies are validated through the various operating conditions of engine experiments. From the experimental results of the robust MIMO feedback design methodology, the design results are satisfied the required tracking performance and robustness as expected. In the case of the second approach, the proposed feedforward controller successfully compensates the non-linear behavior of EGR and VGT systems during transient responses.