Suk-Min Moon

Optimal sensing strategy for adaptive control of optical systems

Optical jitter degrades the pointing and imaging performance of precision optical systems. When a correlated measurement of the disturbance is available, improved control performance can be attained. In this research, an adaptive optimal sensing strategy for optical systems is proposed. An array of reference sensors makes it possible to estimate the disturbance and model the disturbance-to-reference paths. The least-square algorithm is applied for the disturbance model estimation. A sensor scoring algorithm is then used to select an optimal disturbance reference from the available reference signals. The optimal disturbance reference is comprised of sensors which are well correlated with the disturbance. This disturbance reference is then fed forward and used in an adaptive generalized predictive control design. This adaptive control approach is advantageous in the presence of time-varying or uncertain disturbances. The proposed technique is applied to an experimental test bed in which an array of accelerometer sensors measures the structural vibration of optical elements. Reduction of the structural vibration of optical components is attained using a fast steering mirror which results in a reduction of the corresponding jitter. Performance using optimally selected disturbance reference is shown to be better than for system in which a disturbance reference signal is chosen to be the sensor with the lowest score.

ADAPTIVE CONTROL FOR INTERIOR NOISE CONTROL IN ROCKET FAIRINGS

This work describes the development and application of interior noise control techniques based on two adaptive feedback control methodologies. First, an adaptive feedback control technique based on Q-parameterization is used to augment a fixed-gain controller with an adaptive FIR filter for adaptive disturbance estimation. By adjusting the FIR filter coefficients in realtime, the controller is able to adapt to time-varying sound pressure spectrums. Second, a recursive generalized predictive control algorithm, combining the process of system identification and the process of the controller design, is presented for noise control. Identifying the model and designing the controller in realtime enables the controller to fully adapt to time-varying plant dynamics and timevarying disturbances. Experimental results, obtained from a cylindrical enclosure modeled after a launch vehicle, are used to demonstrate the effectiveness of the adaptive Q-parameterized controller and recursive generalized predictive controller. Both control strategies, while very different in implementation, essentially led to the same result, due to the limitations imposed by the physical system. These limitations include the global dynamics of the acoustic space as well as speaker and microphone positions. Details of the design procedures and experimental applications are discussed.

Recursive generalized predictive control for systems with disturbance measurements

The recursive generalized predictive control (RGPC), which combines the process of system identification using recursive least-squares (RLS) algorithm and the process of generalized predictive feedback control design, has been presented and successfully implemented on testbeds. In this research, the RGPC algorithm is extended when the disturbance measurement signal is available for feedforward control. First, the feedback and feedforward RGPC design algorithm is presented when the disturbance is stochastic or random, and is applied to an optical jitter suppression testbed. Second, the feedback and feedforward algorithm is further extended when the disturbance is deterministic or periodic. The deterministic disturbance measurement is used to estimate the future disturbance values that are then used in the control design to enhance the performance. The RGPC with future disturbance estimation algorithm is applied to a structural system and an acoustic system.

Recursive methods for optical jitter suppression using acoustic actuators

The recursive generalized predictive control algorithm, conjugating the process of system identification and the process of the controller design, is presented and applied to real time. In the control design process, there are three parameters to be chosen: the prediction horizon, the control horizon, and the input weighting factor. The prediction horizon and the control horizon are the finite horizons of the system output and control input predictions. Two practical parameters are defined to express effects of the prediction horizon and the control horizon. A time varying algorithm for the input weighting factor and a dual-sampling-rate algorithm are presented. A time varying input weighting factor algorithm allows the recursive generalized predictive controller to be designed aggressively. A dual-sampling-rate algorithm between data acquisition and control design allows higher-order controllers to be designed. The recursive generalized predictive control algorithm is applied to two different systems: a sound enclosure and an optical jitter suppression testbed. For each experiment, the estimation of the frequency response magnitude corresponding to the open loop identified system and the closed loop identified system is shown and compared. The advantages of the proposed recursive generalized predictive control algorithm are: no prior system information is required since the process of the system identification is performed recursively from real time system input and output data, and the controller is updated adaptively in the presence of a changing operating environment.

The Theory and Experiments of Recursive Generalized Predictive Control

The recursive generalized predictive control algorithm, combining the process of system identification and the process of the controller design, is presented. In the control design process, there are three parameters to be chosen: the prediction horizon, the control horizon, and the input weighting factor. Two new parameters are defined for the practical choice of the prediction horizon and the control horizon. A time varying algorithm for the input weighting factor and a dual-sampling-rate algorithm between system identification and control design are presented. The recursive generalized predictive control algorithm is applied to two different systems: a sound enclosure and an optical jitter suppression testbed.Copyright

Real-time optimal sensing strategies for active control of optical systems

The pointing and imaging performance of precision optical systems is degraded by disturbances on the system that create optical jitter. These disturbances can be caused by structural motion of optical components due to vibration sources that (1) originate within the optical system, (2) originate external to the system and are transmitted through the structural path in the environment, and (3) are air-induced vibrations from acoustic noise. Beam control systems can suppress optical jitter, and active control techniques can be used to extend performance by incorporating information from accelerometers, microphones, and other auxiliary sensors. In some applications, offline fixed gain controllers can be used to minimize jitter. However there are many applications in which a real-time adaptive control approach would yield improved optical performance. Often we would like the capability to adapt in real-time to a system which is time-varying or whose disturbances are non-stationary and hard to predict. In the presence of these harsh, ever-changing environments we would like to use every available tool to optimize performance. Improvements in control algorithms are important, but another potentially useful tool is a real-time adaptive control method employing optimal sensing strategies. In this approach, real-time updating of reference sensors is provided to minimize optical jitter. The technique selects an optimal subset of sensors to use as references from an array of possible sensor locations. The optimal, weighted reference sensor set is well correlated with the disturbance and when used with an adaptive control algorithm, results in improved line-of-sight jitter performance with less computational burden compared to a controller which uses multiple reference sensors. The proposed technique is applied to an experimental test bed in which multiple proof-mass actuators generate structural vibrations on a flexible plate. These vibrations are transmitted to an optical mirror mounted on the plate, resulting in optical jitter as measured by a position sensing detector. Accelerometers mounted on the plate are used to form the set of possible optimal reference sensors. Reduction of the structural vibration of optical components is attained using a fast steering mirror which results in a reduction of the corresponding jitter.

Estimating the number of uncorrelated disturbance sources in structural systems

The number of input signals used as reference inputs for feedforward control applications is limited due to cost, computational burden, input processing capability, and installation constraints. Identifying an optimal subset of reference sensors from a larger set capable of conveying the dynamics important in the performance path can result in greater performance with reduced complexity and order in the active control system. However, before determining an appropriate subset of sensors, the number of exogenous disturbance sources must be determined. Principal component analysis can be used to determine the number of uncorrelated disturbances acting on a structural system. Singular value decomposition of a covariance matrix of measured sensor signals is used to determine the number of independent disturbances present in the dynamic system. Limitations imposed by sample data length, path dynamics, and noise can limit the ability to resolve the number of exogenous disturbance sources. To estimate the number ...

Advanced technique to estimate the number of uncorrelated disturbance sources in structural systems

In many control applications the number of input signals for use as reference sensors in feedforward control is restricted due to limitations on cost, computational burden, input processing capability, and cabling and installation issues. Techniques to identify an optimal subset of reference sensors from among a larger set of possible sensor locations is therefore beneficial in attaining improved closed-loop control performance. This optimal subset is a minimum number of sensors that together convey the dynamics important in the performance path. To determine this subset, an approach for estimating the number of exogenous disturbance sources is first required. The proposed technique for estimating the number of uncorrelated disturbance sources acting on a structural system from sampled data is based on the Principal Component Analysis (PCA). This classical statistical method is used in conjunction with Singular Value Decomposition (SVD), and PCA/SVD of a covariance matrix of measured sensor signals is used to determine the number of independent disturbances present in the dynamic system. In practice however, this is difficult to resolve due to limitations on sampled data length, path dynamics, and the influence of noise. The dominant disturbance is evident but secondary disturbances important for the performance path of interest are not readily apparent. In order to better estimate the number of secondary disturbance sources, the addition of a control signal source to minimize the sensor response due to the dominant signal source is proposed. When such a control signal is applied to the system, it is then possible to determine the number of significant secondary signal sources using PCA/SVD. This improvement results in a better determination of the minimum number of reference sensors required for feedforward control of disturbances to structural systems.

Optimal Sensing/Actuation Strategies for Vibration and Acoustic Control of Optical Systems

Optical jitter can result in the beam pointing inaccuracy and poor optical system performance. With a correlated measurement of the disturbance, improved control performance can be achieved. In this research, an adaptive optimal sensing strategy for optical systems is proposed. When an array of reference sensors is available, an optimal set of reference sensors that are coupled to modes of interests can be selected. The weighted reference signal from the optimal sensor set is then used in an adaptive control design algorithm. An adaptive generalized predictive control design algorithm combined with the proposed adaptive optimal sensing strategy achieves better performance than the control system using only one of the reference sensors. The overall algorithm is also advantageous in the presence of time-varying or uncertain disturbances. The proposed technique is applied to an experimental test bed in which multiple accelerometer sensors measure the structural vibration of optical elements. Reduction of the structural vibration of optical components is attained using a fast steering mirror which results in a reduction of the corresponding jitter.

Adaptive generalized predictive control combined with a least-squares lattice filter

The generalized predictive control (GPC) concept is extended to an adaptive control algorithm by combining with a least-squares lattice filter. A least-squares lattice (LSL) filter, another class of exact least-squares filters, has a modular structure that is advantageous in the application of on-line system identification. The modular structure passes system information from lower order to higher order in a wave motion. The adaptive GPC algorithm combined with a LSL filter is implemented for a real-time computer algorithm and its performance is experimentally demonstrated to a structural system and an acoustic enclosure. In addition, the adaptive GPC algorithm with a LSL filter is compared with the adaptive GPC algorithm combined with a classical recursive least-squares (RLS) filter in terms of complexity, computational cost and other on-line application concerns. The average task execution time (TET) --- the measured processing time to run the algorithm during each sample interval --- is reduced by over 35 \% by using the adaptive GPC algorithm with a LSL filter.