Soft-sensor Method For Mechanical Parameters Of Multi-DOF Ultra-precision Motion Stage

As the core part of numerous precision processing equipments, t he performance of a multi-DOF precision motion stage determines the standard of the corresponding products. To simplify the structure of the stage, its actuators are mounted in a highly coupled way, and so are its sensors. Therefore, it is necessary to decouple the measurement system and motor output to control the stage. Many mechanical parameters are contained in the decoupling model. In the machining and assembling processes of the stage, however, machining errors and assembly errors are unavoidable, such as the centre point error and direction deviation of the motor output’s forces, position error of the stage’s centroid, misalignment of the sensors’ input axes, and so on. Therefore, directly applying the theoretical parameters in the actual system will cause decouple errors, and thus performance degradation of the whole system. To achieve precision servo control of the multi-DOF motion stage, parameters in the decouple model have to be accurately determined. Unfortunately, most of these parameters are difficult to be measured directly. Therefore, an indirect soft-sensor method is required to estimate the parameters.This thesis is focused on the reticle stage in a lithography machine, which is representative of sophisticated precision motion stages. The s oft-sensor method for determining the parameters in the decouple models of the driving and measuring systems is studied.First, the configuration of the reticle stage is analyzed, covering its whole architecture, mechanical system, drive system, measurement system and control system, which lays the foundation for follow-up work.Then, the algorithms of soft-sensor are studied. Based on comparison, the heuristic optimization algorithm TLBO is chosen as the core algorithm for the soft-sensor work. And an adaptive hybrid TLBO algorithm is proposed to address the probable premature problem when using the original TLBO. The superiority of the proposed algorithm is verified by Benchmark testing.After that, the soft-sensor method is applied to determine the parameters in the Hall sensor measurement system, whose decouple model and output correction model are derived. The effectiveness of soft-sensor method is validated by testing results from the actual system.Finally, the soft-sensor method is applied to determine the parameters in reticle stage’s driving system, whose decouple matrix is derived. And the simulation result proves its advantage in convergence for the model. In addition, the effectiveness of the proposed method in the planar 3-DOF is validated by experimental results from the actual reticle stage.