Methods of Constructing a Blended Performance Function Suitable for Formation Flight

This paper presents two methods for constructing an approximate performance function of a desired parameter using correlated parameters. The methods are useful when real-time measurements of a desired performance function are not available to applications such as extremum-seeking control systems. The first method approximates an a priori measured or estimated desired performance function by combining real-time measurements of readily available correlated parameters. The parameters are combined using a weighting vector determined from a minimum-squares optimization to form a blended performance function. The blended performance function better matches the desired performance function minimum than single-measurement performance functions. The second method expands upon the first by replacing the a priori data with near-real-time measurements of the desired performance function. The resulting blended performance function weighting vector is updated when measurements of the desired performance function are available. Both methods are applied to data collected during formation-flight-for-drag-reduction flight experiments.