Modeling, analysis, and robust controls for electrophotographic imaging systems subject to spatially periodic disturbances and measurable nonlinearities

Cheng-Lun Chen, Purdue University

Abstract

This thesis is devoted to modeling, identification, control, and performance improvement for a class of electrophotography (EP)-based imaging systems, which are commercially known as laser printers. For this type of systems, halftone banding is a well-known artifact that causes periodic density variation perpendicular to the print direction on printed images. It is known that gear eccentricity and tooth profile error are main contributors to scanline spacing variation, which introduces density variation. Using frequency domain method, we were able to identify the density variation of a certain frequency with the corresponding gear component(s). Furthermore, we developed a model to quantitatively relate printer parameters (optical, electrical, and mechanical) to halftone banding. To further characterize perceived banding, a three-stage approach was proposed, which includes a nonlinear filtering process that emulates human visual system. Preliminary results that support the proposed filtering process were presented. Another focus of the work is to propose and investigate the designs of repetitive control based systems for the image formation system of EP processes. The repetitive controller was successfully incorporated into or combined with the design framework of other linear robust controllers, e.g., H ∞ control. Furthermore, human contrast sensitivity function (CSF) was incorporated as the performance weighting. For sensor and actuator requirement, it was argued that sensors should be placed as close to intermediate images as possible. The number of actuators should at least equal to the number of imaging components. These arguments were supported by experimental and analytical results for a system with non-ideal sensor placement and another with an under-actuated motion subsystem. For the system with an under-actuated motion subsystem, banding reduction was attained by introducing a nonlinear feedforward control by modulating the laser exposure, which complemented the feedback control of the motion subsystem. To preserve the performance of the repetitive control for motion subsystems subject to spatially periodic disturbances, we reformulated the temporal domain system with respect to a spatial coordinate, e.g., photosensitive drum angular position, which resulted in a nonlinear position invariant (NPI) system. Several controller design approaches were suggested. By viewing the angular velocity and the ratio between actuator input and output as varying parameters, the nonlinear system with actuator saturation can be treated as a linear parameter varying (LPV) system, for which an LPV gain-scheduling repetitive controller (LPVRC) with anti-windup design was performed. Experimental results were presented to verify the effectiveness of the LPVRC design.

Degree

Ph.D.

Advisors

Chiu, Purdue University.

Subject Area

Mechanical engineering|Electrical engineering

COinS