| Popis: |
When designing control systems, it is important to keep in mind the end-users. In general, the complexity and difficulty of implementing and tuning a control strategy determine if such a strategy should be adopted, despite its performance. The automotive industry is no exception to that; powertrain control system engineers are constantly tasked to design controllers that are simple to follow, fairly robust, and require moderate resources for implementation and validation. Obviously, stability is critical for any control strategy; an oscillating controller not only could result in fluctuation in engine torque which could lead to customer dissatisfaction, but also could have a negative impact on vehicle tailpipe emissions. Bottom line is the controller must provide good stability as well as superb reference tracking. Proportional–Integral-Derivative (PID) controllers are attractive solutions for powertrain calibrators. They are widely used in the automotive industry as they are easy to implement and tune, especially when the Derivative term is dropped to reduce the sensitivity of the system and improve its robustness. Consequently, with only two gains to adjust, tuning a PI controller is a relatively simple task, but quite often calibrators struggle trying to meet both stability requirements and tracking performance. Adding a Smith Predictor (SP) to the PI controller, is one way to achieve the tracking performance desired. This is indeed a well known approach that allows for much higher gains to be used in order to improve the tracking ability while maintaining stability in the presence of plant models changing dynamics and system delays. However, with this combination (PI-SP), the robustness of the overall control strategy suffers as errors in the plant models tend to reduce the stability of the over all system and could potentially lead to an unstable or undesirable performance. In this paper, we will present a practical approach to improve the performance and robustness of the PI controller. This is achieved by adding unity-gain feed-forward (FF) of the air fuel set point for improved tracking plus a set-point filter in the input to the PI controller (FF-PI) to prevent overshoots. We will also show that this control structure provides excellent and robust reference tracking without destabilizing the system in the presence of errors in the plant model. In addition, we will include results obtained with the FF-PI controller when applied to two applications: Air-fuel ratio (A/F) control for gasoline lean burn vehicles (5.4L F150 and 4.0L Ford of Australia Falcon), and Exhaust Gas Recirculation (EGR) control on a 2.0L Gasoline, Turbocharged, Direct Injection Engine. |
