Light Years Closer

Nonlinear filter increases telescope stability and pointing accuracy

By Monty J. Smith, Ph.D. and David C. Slater, Ph.D.

Dr. Monty Smith is a research engineer in the Engine and Vehicle Research Division. His responsibilities include advanced developmental controls research and modeling for the aerospace, automotive, electronic and manufacturing industries. Areas of expertise also include hardware development and implementation for new and enabling technologies in the control, identification and signal processing arenas using the latest in processor technology.

Scientists looking into deep space through the lens of time continuously strive for more accurate instrumentation to enhance their vision of the universe. Random noise, either electronic or structural, impairs the ability of these instruments to gather accurate readings, a problem that has been previously addressed through linear filtering technology. An internally funded Southwest Research Institute project has shown, however, that even greater accuracy is possible through the use of nonlinear filters.

Nonlinear filter technology, recently developed at SwRI for improved telescope pointing, helps resolve problems associated with power sensitivity and alignment errors in instruments used to study asteroids, comets, planets and other space objects. This technology also has applications for missile guidance, aircraft flight control systems, industrial processes and improved machine tool accuracy. For example, this technology might have improved the Gulf War performance of the U.S.-launched Patriot missile by keeping it on an ultra-high precision trajectory. Studies performed after the 1991 Gulf War have shown that Patriot missiles destroyed with high confidence only 25 percent of the enemy's Scud missile warheads.

Electronic and structural noise, inadvertently introduced into all telescopic hardware, adversely affects instrument precision, alignment and pointing. Structural noise includes vibration of flexible space structures caused by re-alignment maneuvers and contraction and expansion of material from exposure to extreme temperature changes. Electronic noise can originate from transformers, power supplies and instrumentation. Devices relying on digital information for processing and sending corrective signals to instrument hardware can be installed to correct or attenuate deviations caused by noise. Their digital logic is derived from linear systems theory because of computational restrictions imposed by existing computer platforms.

Motivated by the progression of core processor technology, SwRI researchers proposed the new, nonlinear filtering approach with applications directed toward the improvement of astronomical hardware. Additional numerical simulation based on earlier theoretical studies gave SwRI investigators confidence that instrument pointing accuracy could be improved by an order of magnitude.

Background

Linear filters to control the effects of noise in instruments used to survey the heavens were developed long before the advent of the digital computer. Some of the earliest analytical contributions came from Bell Laboratories for the design of electronic amplifier circuits during the 1930s. These filters also estimate problems caused by random noise and reconstruct the signal. Engineers working to resolve disturbance attenuation issues often implement linear filters, typically Kolmogorov Weiner and Kalman filters, on computers using closed form solutions, or solutions based upon a fixed number of computations between sampling intervals. With recent technological advances in digital signal processing (DSP), instrumentation has become readily available that enables engineers to implement nonlinear, digital filtering strategies with performance that was previously unachievable.

Conventional Filter	Low Power Filter
	The grimbal angular position error and motor command voltage signals were recorded using the low power, conventional, and maximum accuracy filters under similar noise environments. The additional freedom in nonlinear filter design accounts for predetermined power constraints or results in improved instrument precision.
Maximum Accuracy Filter

Experimental platform

To test the hardware, SwRI researchers developed nonlinear filtering algorithms with a gimbaled, laser-pointing experiment that emulates a telescope. Essentially, these nonlinear filtering strategies give the instrument designer two options beyond the conventional approach based upon the following criteria: low power (with stability preserved) and maximum accuracy (using all available power).

SwRI staff compared variations of low power, maximum accuracy nonlinear filters to the conventional filtering methodologies being used today. Researchers chose a Texas Instruments evaluation module based on DSP technologies for the final platform for successful implementation of these nonlinear filtering strategies.

Experimental results

To test the capabilities of nonlinear filtering for noise rejection, SwRI staff injected random electronic noise into the hardware to induce unwanted excitation. They used three filters -- conventional, low power and maximum accuracy -- under the same noise environment for comparative studies.

Researchers implemented all three filtering strategies under the influence of artificially generated external noise. The latter two approaches involved the implementation of nonlinear filtering techniques developed at SwRI.

The experimental stage of this program readily proved the effectiveness of using nonlinear filtering to reduce the effects of exogenous disturbances. Using the maximum accuracy filter, SwRI engineers observed from the data that a 20 percent decrease in error over the conventional approach could be achieved under the same disturbance/noise environment. In addition, with stability as the primary objective and performance the low priority, the implementation of the low power filter decreased energy requirements by 74 percent.

Conclusion

SwRI is developing an extension to this nonlinear filtering approach. Other organizations have expressed interest in this technology, and communication is under way for furthering the development for their specific applications.

Comments about this article? Contact Smith at (210) 522-3208 or msmith@swri.org.

Published in the Fall 2001 issue of Technology Today®, published by Southwest Research Institute. For more information, contact Maria Martinez.

Technics Fall 2001 Technology Today
SwRI Publications SwRI Home