The design of closed-loop controllers for high precision servo positioning systems in high-volume production is severely challenged by the tradeoff between the controller's robustness and performance. This presentation will review an industrial control design process for a population of dual-stage hard disk drive servo actuators. A multi-rate digital controller is developed to suppress high-frequency resonant modes and track low-frequency desired trajectories. The controller includes a proportional-integral controller cascaded with a set of notch and peak filters. The controller parameters are tuned via an optimization process based on a number of closed-loop system objectives such as stability margins and bandwidth, directly applied to the plants' frequency response data. To further enhance the performance of the closed-loop controller, the plant population is clustered into groups of similar plants, and separate controllers are designed for each cluster. Results indicate considerable improvements in the robustness and performance of the system using the clustering-based control design method.
For long-range motions, the controllers can be supplemented with polynomial reference and feedforward input trajectories for fast and smooth transition from the initial to the target position. A new polynomial design process will be presented, which reduces the transient vibrations in long-range motion by incorporating the flexibility of the system in the reference trajectory. Simulations and experimental tests indicate significant improvements in the settling time of flexible servo systems using the proposed polynomial design framework.
Speaker(s): Dr. Saeid Bashsah
Agenda:
6:00 – 6:30 PM : Networking – Light dinner
6:30 – 7:30 PM : Talk and Q&A
7:30 – 8:00 PM : Networking
Room: SCDI – 2301, Bldg: Sobrato Campus for Discovery and Innovation – SCDI, Santa Clara University, 500 El Camino Real , Santa Clara, California, United States, 95053
Data-Driven Control Design for High Precision Servo Positioning Systems in High Volume Production
- This event has passed.
