Peak-seeking algorithms optimize physical processes in real time and are widely used throughout a variety of industries. However, measuring associated parameters in changing conditions, and responding to them appropriately, is difficult because the measurements are typically distorted by noise. This technology addresses that problem by employing a time-varying Kalman filter. The filter is an algorithm previously developed to estimate unknown parameters within systems that are intrinsically random and uncertain. The Kalman filter is excellent at finding estimates when it encounters noisy signals. As a minimal-variant filter, it inherently produces the best estimates of a function with the smallest amount of variation from the true value. Thus, the filter can help accurately determine the optimal coordinates of the peak- seeking function as conditions in the environment change.
A peak-seeking algorithm can optimize performance of complex operations in real time. Originally designed for aircraft flying in formation, the algorithm automatically finds optimal formation configurations to reduce aircraft drag and therefore increase fuel efficiency. The method is capable of using real-time measurements and quickly adapting to changing environmental conditions.
In addition, this technology takes a multiple-input and multiple-output approach to design for all dimensions simultaneously. This is in contrast to traditional peak-seeking architectures, which can only be designed by considering one dimension at a time, making the design effort difficult. Typical peak-seeking architectures also require specific persistent excitation signals; the new method allows nearly any persistent excitation signal to be used, allowing for greater flexibility.
This approach overcomes the uncertain nature of wake vortex upwash fields with a real-time, onboard solution. It also improves wing efficiency in formation flight without requiring a wing redesign, enabling use of existing control surfaces. Implementation of the algorithm can increase drag reduction by an additional 2 to 3% over formation-flight performance with traditional roll trim schedules. In addition, it enhances formation-flight-for-drag-reduction without sacrificing performance during solo flight.
The peak-seeking algorithm has also been used to optimize an airplane’s performance. The drag of an aircraft is reduced by adaptively determining the optimal trim configuration of the aircraft’s effectors based on real-time in-flight measurements as flight conditions change.
Use of the algorithm has the potential to shorten development time and improve flight test efficiency. It also allows for optimization of multiple aircraft effectors simultaneously. Implementation can result in significant cost savings associated with the reduction in aircraft fuel consumption, which will also lower aircraft emissions and reduce environmental impacts.
This work was done by John Ryan of Armstrong Flight Research Center in collaboration with University of California, Los Angeles professor Jason Speyer. DRC-009-026