Research On Parameter Self-learning Supported Multi-variable Cooperative Control Of Diesel Engine Air System Equipped With VGT And EGR

Using variable geometry turbocharger(VGT)and exhaust gas recirculation(EGR)technologies to control the boost pressure and EGR rate of diesel engine air system is significant to improve the power,economy and emission characteristics of diesel engine.However,the air system of diesel engine is a typical multivariable and strongly coupled complex nonlinear system,and its control is faced with engineering problems such as complex calibration and difficult parameter adjustment.In view of the problems and situations above,this paper proposes a multivariable collaborative decoupling control algorithm for air system under backstepping structure,together with the parameter selflearning function for the algorithm.This algorithm not only realize the decoupling of air system and achieve the ideal control effect,but also reduce the controller’s complexity and burden of calibration.Firstly,this paper takes a six-cylinder four-stroke VGT-EGR diesel engine as the research object,and establishes a simulation platform based on GT-Power.In order to achieve the required accuracy,the object model of the simulation platform is calibrated through the test data from engine bench tests.In order to verify the effectiveness and engineering realization of the algorithm in the embedded controller,hardware-in-loop(HIL)test platform is built.Secondly,the decoupling control algorithm of the air system is designed to solve the difficulty of air system control.A control-oriented air system model is established to serve control algorithm’s design.It is proved that the accuracy of this controloriented model can reach over 85%.Decoupling of air system is realized based on backstepping control structure and multiple-input multiple-output active disturbance rejection control(MIMO ADRC).The multivariable extended state observer(ESO)and the linear state error feedback control law are designed to complete the decoupling control of pre-turbine pressure and EGR rate in the air system.Pre-turbine pressure is an important coupling point of the air system.Based on backstepping,the supervision of pre-turbine pressure and the control of boost pressure are realized,so that the collaborative control of the boost pressure and EGR rate of the air system is completed.The algorithm introduces model information feedforward and target value change rate feedforward to further modify the control law,which can also improve the control accuracy and optimize the dynamic response characteristics.Proportion-integrationdifferentiation(PID)class,model information class and disturbance rejection class are three typical air system control paradigms.According to these paradigms and air system control-oriented model,three kinds of contrast controllers are designed for the comparative verification of this study.Thirdly,to solve the tuning problem of control algorithm parameters,based on the controlled object information and recursive least squares(RLS),the algorithm parameters are self-learned.To ensure the accuracy of learning,a three-tier management framework of " persistence of excitation judgment-parameter rationalityparameter closed-loop control effectiveness" is designed to supervise and manage the parameter self-learning process,so as to ensure that the learned parameters are reasonable and accurate.The verification results of the parameter self-learning module show that with the application of the management framework,the self-learning algorithm converges quickly and the learning effect is greatly improved.Moreover,the results verify the effectiveness and robustness of the module under full operating conditions.Finally,model-in-loop(MIL)test,co-simulation test,HIL test and corresponding comparison verification of the algorithm are carried out.MIL test results show that the backstepping structure,MIMO control mode and the use of model information feedforward can effectively cooperate with actuators,improve dynamic response speed and reduce response overshoot.The use of target value change rate feedforward can eliminate the tracking phase error effectively.Compared with contrast controllers based on traditional algorithms,the controller proposed in this paper shows obvious advantages in control adjustment time and overshoot.Co-simulation test results verify the robustness and superiority of the proposed algorithm in all operating conditions.In the FTP75 driving cycle test,using the proposed algorithm,the integral absolute error(IAE)value of boost pressure and EGR rate are 1.82% and 2.99%,54% and 57% lower than those of the contrast control algorithm,respectively.Processor-in-loop(PIL)test results on Aurix TC397 verify the effectiveness and real-time performance of the proposed algorithm in embedded hardware system.HIL test results show that the proposed algorithm can realize fast,stable and accurate control of the air system under full operating conditions.The control system is realizable and effective on software and hardware devices such as d SPACE,which verifies the feasibility,effectiveness and robustness of the algorithm applied to actual controlled objects.To sum up,the air system decoupling control algorithm proposed in this paper has ideal performance and robustness,solving the difficulties in calibration and parameter adjustment.Also,it is easy to realize in engineering.