| The objective of this research was to develop control structures and test models for automating the bending of straight steel reinforcing bars. Three major thrusts provided the focus for approaching this problem: (1) springback prediction using real-time electronic sensing, (2) adaptive control of the bending operation, and (3) transfer of the control system from a laboratory prototype bender to a shop table bender.; A control framework is presented for linking control of rebar fabrication automation to planning, scheduling, and control in a CIC environment. A basic theory for evaluating the cost of quality through automation control information at the task level is described.; Bending tests were conducted with a shop table bender to establish and compare alternative models for springback prediction. Comparisons of the performance of the alternative models demonstrated the feasibility of successful controller operation. A neural network springback prediction model was included in the comparative study.; The bending torque curve is an appropriate model for monitoring, and controlling the operation of a table bender. Components from the laboratory prototype table bender control system were successfully transferred to the shop bender. Critical components of the control structure were adapted to the new machine through substitution of sensors and creation of new control laws.; Springback model evaluations revealed that empirical statistical models, neural networks, and in-process relaxation perform well. Nearly 95 percent compliance with an extrapolated industry standard is achieved with the existing experimental testing facility. Shared control through a human-machine interface, however, may be the best alternative for achieving highest quality standards and improving worker performance in the areas safety and productivity. |
