Data-driven Air Path Predictive Control Of Turbocharged Gasoline Engines

With the continuous improvement of quality of life, people’s requirements of the car’s performance are also rising, such as good comfort, good power and fuel economy. Against this background, exhaust turbocharged engine emerges. Compared with the naturally aspirated engine, the exhaust turbocharged engine can use the exhaust gas as the power source to compress the intake air, so that the fuel can burn more fully, and the output torque can be improved, and thus the exhaust turbocharged engine has gradually become the mainstream product in the engine.The air path of the exhaust turbocharged engine has two actuators, throttle and wastegate. The two actuators are interrelated and interacted with each other. This characteristic brings challenge to realize the coordinated control the exhaust turbocharged engine. At present, domestic and foreign countries have done a lot of research on this issue, and have achieved some results, such as, feedback linearization control, IMC control.However, these control algorithms rely on high explicit mathematical model, and the air path of turbocharged engine is complex and contains many maps, so that it is very difficult to obtain an explicit model.Taking the modeling difficulty into consideration. This paper design a data-driven based model predict control algorithm, and proposes the following control scheme at the premise of taking the torque demand as the center: Firstly, the desired output torque is transformed into the desired the cylinder air mass flow by looking up a map; Then,data-driven based model predict control algorithm is used to track the the cylinder air mass flow. It is important to select the right inspired data for designing the data-driven based model predict controller. This paper selects throttle opening, wastegate opening and engine speed as the inspired input data, and selects intake manifold pressure, exhaust manifold pressure and cylinder air mass flow as the inspired output data through detail analysis of turbocharged gasoline engine air path. In the controller design process, firstly,Hankel matrix is established by the inspired data, and the predictive equation is got by the least squares method, and then the predictive equation is separated into three independent predictive equations. The dimension of the matrix is decreased in this way, and the amount of calculation of the controller is reduced at the same time. Secondly, the cost function is designed, which can track the cylinder air mass flow, and also limit the intake/exhaust manifold pressure radio and control actions, and then the particle swarm optimization algorithm(PSO) is used to solve the cost function. Finally, the controller is verified in the engine software en DYNA and the Rapid Prototyping Control test bench which is based on d SPACE and x PC-Target. The simulation results show that the controller has good control performance.