Feng-Chi Hsieh

Sideslip angle estimation using extended Kalman filter

Vehicle sideslip angle can be estimated using either dynamic or kinematic models. The dynamic model requires vehicle parameters, which might have uncertainties due to different load conditions, vehicle motions, and road frictions. Parameter uncertainties might result in estimation errors. Thus system identifications are required to estimate those parameters online. On the other hand, the kinematic model does not require these parameters. A closed-loop estimator can be formulated to estimate the sideslip angle using the kinematic model. Since the system matrix which consists of the yaw rate is time varying, the required input vector and output contain process and measurement noises, respectively, and the disturbance input matrix contains estimated states, extended Kalman filter is used to obtain the estimation gain in this paper. CarSim is used to evaluate the proposed approach under different driving scenarios and road frictions in Matlab/Simulink. The preliminary results show promising improvement of the sideslip angle estimation.

Sliding-mode control for semi-active suspension with actuator dynamics

A sliding-mode controller (SMC) is proposed for semi-active suspensions to achieve ride comfort and handling performance simultaneously. First, a nonlinear quarter-car model of Macpherson strut suspension is established in Matlab/Simulink. Constrained damper force and actuator dynamics are considered for the damper model. System identification is applied to the nonlinear model for obtaining the linear model parameters. Kalman filter is designed based on the linear model and the actuator dynamics to estimate the state responses required for SMC. The sliding surface consists of tyre deflection and sprung mass acceleration. The proposed SMC is evaluated using the nonlinear model for both time and frequency domain responses. Robustness due to the increased sprung mass and deteriorated suspension is also investigated in this paper. Preliminary simulation results show improved ride comfort without sacrificing the road holding performance.

Adaptive Idle Speed Control for Spark-Ignition Engines

Due to the nonlinear time-varying nature of the sparkignition engine, an adaptive multi-input single-output (MISO) controller based on self-tuning regulator (STR) is proposed for idle speed control in this paper. The spark timing and idle air control are simultaneously employed as control inputs for maintaining the desired idle speed, and are designed based on P and PI type STR, respectively. The Recursive Least Square technique is employed to identify the engine as a first-order MISO linear model. Pole placement technique is then used to design the adaptive MISO controller. Performances of the proposed algorithm are evaluated using a nonlinear engine model in Matlab/Simulink. The system parameters with 10% uncertainties are also utilized to perform the associated robustness analysis. Preliminary simulation results show significant reduction of speed deviations under the presence of torque disturbances and model uncertainties.

Modulization of four-stroke single-cylinder spark-ignition air-cooled engine models

Abstract In order to satisfy different requirements for engine design and real-time simulation, modulization technology is used in this paper to establish the engine model for small-scale engines. The model consists of simple and complex modules of charging, torque, friction, and crankshaft dynamics, which are established in Matlab/Simulink and verified using the experimental data. Different sets of these modules can be selected for various applications. For engine design, a complex model, which consists of the wave-action charging module and the mean-value combustion module, is employed to study the effects of inlet and exhaust systems on torque output performance. For real-time simulation, different levels of complexity can be selected according to the hardware-in-the-loop requirement of the control verification.

Design of Multi-Mode Switch Strategy for Lean Burn Engine Using Driving Pattern Recognition Technique

A multi-mode switch strategy based on driving pattern recognition (DPR) technique used for lean burn engine is proposed in this paper. First, four representative driving patterns (RDP) are selected from nine Taiwan driving patterns. The single-mode switch strategy is then extracted and optimized for each RDP by analyzing the results of dynamic programming. The proposed strategy can select the appropriate single-mode switch strategy using the DPR technique, which is designed to classify the current driving scenario into one of the RDPs. Preliminary results show that the proposed strategy can significantly reduce the fuel consumption without excessive increment of NOx emission across all the test scenarios.

Estimation of engine rotational dynamics using a closed-loop estimator with stroke identification for engine management systems

Abstract In order to study the effect of the crank sensor noise on the engine management system (EMS), an algorithm using a closed-loop estimator with stroke identification is proposed to estimate the engine rotational dynamics. Estimated crank angle and engine speed are used for fuel injection and ignition control systems. The closed-loop estimator design is based on a linear model by assuming that the engine rotational inertia is constant. Since the effective inertia actually varies with different crank angles, the stability of the proposed algorithm is assessed using the Lyapunov stability theorem. Performances of the proposed and traditional algorithms are evaluated using a non-linear engine model with a four-plus-one-tooth crankshaft wheel in Matlab/Simulink. The estimated crank angle and engine speed of the traditional algorithm can be significantly affected by large sensor noises resulting from the poorly grounded ignition coil. It was found that the proposed algorithm can mitigate the noise impact and thus maintain the desired engine control performance.

Estimation of Engine Rotational Dynamics Using Kalman Filter Based on a Kinematic Model

For a conventional scooter engine with a four-plus-one-tooth crankshaft wheel, not only is the crankshaft position estimation insufficient due to poor angle resolution, but the speed measurement might also be easily contaminated by the sensor noise. We proposed a Kalman filter with stroke identification to estimate the engine rotational dynamics. The design of the Kalman filter is based on a kinematic model that requires no engine parameters. A nonlinear engine model is used to evaluate the estimation performance of the conventional algorithm using a low-pass filter and the proposed algorithm at various operating conditions. Preliminary simulation and experimental results show that the proposed algorithm can mitigate the noise impact and result in estimations closer to the actual engine responses.

Oil Coking Prevention Using Electric Water Pump for Turbo-Charge Spark-Ignition Engines

Turbocharger has been widely implemented for internal combustion engine to increase an engines power output and reduce fuel consumption. However, its operating temperature would rise to 340°C when engine stalls. This higher temperature may results in bearing wear, run-out, and stick, due to oil coking and insufficient lubrication. In order to overcome these problems, this paper employs Electric Water Pump (EWP) to supply cool liquid to turbocharger actively when the engine stalls. The system layout, operating timing, and duration of EWP are investigated for obtaining optimal performance. The primarily experimental results show that the proposed layout and control strategy have a lower temperature of 100°C than the conventional temperature 225°C.

Engine Modeling With Inlet and Exhaust Wave Action for Real Time Control

A motorcycle engine model in MATLAB/SIMULINK is introduced in this paper for engine design and control system design. It simulates a 125 cc single cylinder four-stroke spark ignition engine. Sub-models include: inlet/exhaust, combustion, heat transfer, friction, and work done. The method of characteristics is used for calculating the inlet and exhaust flow. Simulated results of bmep, imep, fmep, heat release rate, inlet/exhaust pressure, and cylinder pressure were compared with the experimental data. It was found that satisfactory simulation results were achieved for the proposed model.© 2003 ASME