| At present,it is difficult to control the forming accuracy in the roll bending process of aluminum alloy profiles,which is primarily due to two reasons,First,the roll-bending process of aluminum alloy profiles is a process of locally constrained elastic-plastic deformation,which is prone to springback;Second,the roll-bending process is a process of dynamically changing the position of the profile,and the roll-bending forming size is the size of the profile after bending and rebounding,which is not easy to measure accurately online.Therefore,the roll bending process is mostly open-loop control,and it is difficult to control the forming accuracy.In order to solve the problem of online roll bending accuracy control of profiles,This paper proposes a feed-forward closed-loop control method for roll-bending based on online dimension detection and machine learning,the online closed-loop precision control of the dimensions of rolled-bending parts is realized by online dimension detection and feed-forward prediction based on bending error data.The main research work in this paper is as follows:(1)The roll-bending drive mechanism is designed,based on the working principle of the roll-bending drive mechanism,the dynamic model of the roll-bending drive mechanism and the PID control model of the position loop are established,and the PID parameters are set,which provides the basis for the roll-bending closed-loop control.(2)A feedforward predictive control strategy based on support vector machine is proposed,and a forming curvature prediction model and three error compensation methods based on mold-mounting error compensation,curvature error compensation and radius error compensation are established based on support vector machine,the simulation of the error-compensated profile variable-curvature roll-bending predictive control model provides a basis for the selection of the roll-bending error compensation method.(3)The off-line predictive control experiment of roll bending,the simulation experiment of constant curvature based on the compensation of mold fitting error,curvature error compensation and radius error compensation,and the pre-bending experiment of variable curvature based on curvature error compensation are carried out.Based on the variable-curvature pre-bending experimental data,the variable-curvature roll-bending experiment based on off-line error compensation prediction is realized.(4)The general framework of the roll-bending control system was constructed,the construction of the online predictive control rolling-bending test bench was completed,and the variable-curvature online predictive-control roll-bending experiment was carried out to verify the feasibility of the online predictive control of the profile rolling accuracy. |