SENSOR BIASES EFFECT ON THE ESTIMATION ALGORITHM FOR PERFORMANCE-SEEKING CONTROLLERS

The performance-seeking-control algorithm (PSC) is designed to continu ously optimize the performance of propulsion systems. The PSC uses a n ominal model of the propulsion system and estimates, in night, the eng ine deviation parameters (EDPs) characterizing engine deviations with respect to nominal conditions. In practice, the measurement biases (or model uncertainties) may prevent the estimated EDPs from reflecting t he engine's actual off-nominal condition. This factor has a direct imp act on the PSC scheme exacerbated by the open-loop character of the al gorithm. An observability analysis shows that the biases cannot be est imated together with the EDPs. Moreover, biases and EDPs turn out to h ave equivalent effects on the measurements, leaving it undecided wheth er the estimated EDPs represent the actual engine deviation or whether they simply reflect the measurement biases. In this article, the effe cts produced by unknown measurement biases over the estimation algorit hm are evaluated. This evaluation allows for identification of the mos t critical measurements for application of the PSC algorithm to an F10 0 engine.