Research On Application Of Generalized Predictive Control In Lateral Control Of Intelligent Vehicles

Cars are becoming more and more important in people’s lives,and the traffic problems they are causing are becoming more and more serious.As an important part of intelligent transportation,intelligent vehicles have a low traffic accident rate and more efficient road traffic,which has attracted the attention of scholars at home and abroad.Lateral tracking control is one of the key technologies for studying the movement process of intelligent vehicles,but the lateral control system has obvious coupling output.The generalized predictive control can use known information to repeatedly scroll optimization in a limited period of time,and its integral function can well eliminate steady-state errors,with anti-interference and robustness,and its predictive thinking and Drivers’ driving thinking is similar,and it is a dynamic control.Therefore,in this paper,the generalized predictive controller is applied to the lateral control of intelligent vehicles,and the controller is specifically improved.The details are as follows:(1)A multivariable generalized predictive output decoupling controller for quickly solving the Diophantine equation is designed for a multivariable system with two degree of freedom single track lateral control dynamics model output coupling.First,the control parameters related to the coupling in the output coupling variable matrix are separated and redefined as the expression form of diagonal matrix and anti-corner matrix.Secondly,the controlled autoregressive integral moving average model is split into two related models with and without coupling,and then combined with the improved Diophantine equation to obtain a multivariate generalized predictive output decoupling controller.Finally,the vehicle lateral control model is discretized and converted into a transfer function matrix expression to meet the model requirements of the multivariable generalized predictive output decoupling controller.(2)Aiming at the model output coupling phenomenon,a generalized predictive control optimal performance index function is designed.The output coupling parameter is introduced into the performance index function to eliminate the influence of the output coupling in the lateral control model on the vehicle motion state.First,a matrix polynomial transformation with zero diagonal is performed on the coupling parameters.Secondly,by solving the optimal performance index function,the two parts that are associated with the coupling change and the uncorrelated are separated,the whole part associated with the coupling change is zero,and the introduced output coupling parameters are solved.Finally,the optimal control input and optimal prediction output under the optimal performance index are obtained.(3)Aiming at the single-input and double-output lateral dynamic model of intelligent vehicles,the matrix Diophantine polynomial equation is improved,the matrix Diophantine equation is deformed,the calculation amount is simplified,and the real-time performance of the generalized predictive controller is improved.First,the matrix Diophantine equation is separated from the corresponding coupling terms related to the designed output decoupling controller,and converted into three matrix Diophantine equations.Then,through the correspondence between the elements of the matrix,the matrix is transformed into a polynomial equation with only a single parameter for solution.Finally,by means of matrix transformation,the calculated data are re-arranged in the corresponding positions of the matrix parameters to form a new matrix,that is,the matrix solution of the matrix Diophantine equation is obtained.At last,the improved GPC,GPC and PID controllers are compared and studied on the lateral control of intelligent vehicles,and the effectiveness and superiority of the improved generalized predictive controller are verified.