On-Line System Identification and Control Strategies for Improving the Performance of Inertial MEMS

智能优点：微型机电系统(MEMS)的最新发展导致了各种微型传感器和执行器，它们在飞机和航天器的制导和控制方面具有巨大的潜力。这类设备可以提供高带宽性能和低功耗，并且它们的小尺寸和重量允许在传统的传感器/执行器技术限制测量通道数量和执行器自由度的应用中使用大量的传感器和/或执行器。为了使这些新兴技术实现具有竞争力的性能水平，需要创新的方法来进行设备校准和闭环控制。这种方法的必要性源于这样一个事实，即由于关键机电部件的微观尺度，制造不规则性可以在任何一批假定相同的器件中产生广泛的动态响应。传感器和执行器动力学的灵敏度和变化性可以通过基于输入/输出数据的详细建模和参数辨识的单个器件的精确校准，以及通过内置在微型器件中的控制回路来补偿。广泛的影响：本研究的重点是开发多变量系统辨识和控制策略，以提高MEMS振动角速度传感器的性能。这项研究的学术价值源于对以下问题的研究：1)第一个主要研究课题开发了实现最佳器件灵敏度的算法。当传感器的两个动态模式退化时，可获得最高的信噪比。稳健的测试方法和算法已经被开发出来，通过传感器动态的区域时间调整，极大地提高了与角速度估计相关的信噪比。此外，在线算法在运行过程中不断探测感测动态，以保持调谐条件；2)第二个主要研究主题开发了一种通过整形闭环噪声谱来最小化指定积分时间的速率不确定性的过程。在这种情况下，控制滤波器中的额外自由度被用来以所需的方式形成速率估计的Allan方差。这项研究计划是在PIS微传感器实验室进行的，传感器原型是从NASAS喷气推进实验室借给PI的。这些研究贡献的影响远远超出了研究生教育。在这个项目下设计的算法和方法改变了喷气推进实验室和其他地方的PI合作者建模、测试和控制传感器原型的方式。在加州大学洛杉矶分校的本科工程课程中，一系列设计课程让学生接触到这种快速发展的传感器技术，项目包括使用BEI Technologies捐赠的商用陀螺芯片振动器为加州大学洛杉矶分校DARPA大挑战团队设计和建造稳定的传感器平台。

Intellectual Merit:Recent developments in microelectromechanical systems (MEMS) have lead to a variety of microsensorsand actuators that have great potential for guidance and control of aircraft and spacecraft. Such devicescan deliver high-bandwidth performance with low power consumption, and their small size and weightpermit the use of numerous sensors and/or actuators in applications where conventional sensor/actuatortechnology limits the number of measurement channels and actuator degrees of freedom.For these emerging technologies to realize competitive performance levels, innovative methods are requiredfor device calibration and closed-loop control. The necessity of such methods arise from the factthat fabrication irregularities can produce a wide range of dynamic responses among any batch of supposedlyidentical devices due to the microscopic scale of key electro-mechanical components. The sensitivityand variability of sensor and actuator dynamics can be compensated by precise calibration of individualdevices based on detailed modeling and parameter identification from input/output data, and by controlloops built into the micro device.Broader Impact:This research focuses on the development of multivariable system identification and control strategiesfor improving the performance of MEMS vibratory angular rate sensors. The intellectual merit of thisresearch derives from the study of problems stimulated by the following needs: 1) The first major researchtopic develops algorithms for achieving optimum device sensitivity. The highest signal-to-noise ratiois achieved when two dynamic modes of the sensor are rendered degenerate. Robust test methods andalgorithms have been developed that vastly improve the SNR associated with the angular rate estimate viareal-time tuning of the sensor dynamics. Furthermore, on-line algorithms continuously probe the sensordynamics during operation in order to maintain the tuned condition; 2) The second major research topicdevelops a process for minimizing the rate uncertainty at designated integration times via the shaping ofclosed-loop noise spectra. In this case, additional freedom in the control filters is exploited for shaping theAllan variance of the rate estimate in a desired manner.This research program is conducted in the PIs Microsensor Laboratory on sensor prototypes loanedto the PI from NASAs Jet Propulsion Laboratory. The research contributions have significant broaderimpacts well beyond graduate education. The algorithms and methods devised under this program havechanged the way the PIs collaborators at JPL and elsewhere model, test and control their sensor prototypes.In the undergraduate engineering curriculum at UCLA, a series of design courses expose students to thisrapidly evolving sensor technology with projects that include the design and construction of a stabilizedsensor platform for the UCLA DARPA Grand Challenge Team using commercial GyroChip vibratorygyros donated by BEI Technologies.