Bp Neural Network-based Tension Control System

In the printing and packing industry, gravure printing is widely used in many printing fields because of plenty of merits, such as thick and firm printing ink, bright and shining colors, high flexibility and so on. Control precision of the tension of the pressed webs is one of key techniques to ensure products' qualities. This paper makes a deep and systematical research on tension control in printing machineries, taking gravure printing machine as an object.The paper designs the distributed control system which takes unwinding roller, moment motor, DC motor, traction roller as executors, load cells as detectors, PLCs of 90-series produced by GE as controller. At the basis of configuration of GENIUS network, the system applies WinCC software to develop the monitoring and controlling interface, OPC technology to exchange data between WinCC and PLC, so as to extend the system's compatibility and interoperation.Based on Hooke's Law and the theory of Dynamic Moment Balance, the paper respectively establishes the mechanical and electrical relations of the subjects, including unwinding roller, moment motor, DC motor, load cell, traction roller and so on, and presents the mathematical model of the whole product line of the gravure printing machine in the form of transfer function block diagram. Aiming at the characteristic of printing machine that the system is multi-input, time-variant and nonlinear, it puts forward the tension control system based on BP Neural Network, and simulates the system established on the platform of the Matlab software. The results simulated indicate that the method presented in the paper overcomes the weakness of traditional PID control, and is capable of keeping steady tension in printing process. The fact that dynamic response and control precision has been improved validates the feasibility of the BP Neural Network algorithm.