Continuous optimal infeed control for cylindrical plunge grinding, part 2: Controller design and implementation

This is the second of two papers concerned with on-line optimization of cylindrical plunge grinding cycles with continuously varying infeed control. In the first paper (1), dynamic programming was applied to a simulation of the cylindrical grinding process' in order to explore the characteristics of optimal grinding cycles. Optimal cycles were found to consist of distinct segments each with predominant constraints. An optimal control policy was formulated with the infeed rate within each segment determined according to the prevailing constraint. The present paper is concerned with the design of the controller and its implementation. The control system to implement the optimization policy is described together with provisions to enhance robustness to modeling uncertainty and measurement noise. Robustness provisions include model adaptation by parameter estimation from on-line measurements of size and power, and incorporation of safety margins in the optimization process. Problems associated with practical implementation of the control system, stemming from power limitations and wheel wear, are also discussed. The controller performance is demonstrated on an instrumented internal cylindrical grinding machine.