Linear Regression with Intercept

2006 American Control Conference

2006

Indirect adaptive control for reconfigurable flight control is advocated. Specifically, on-line static system identification using a moving window/batch estimation, is used. The ensuing linear regression is augmented with an intercept to address the effects of trim change associated with the occurrence of a control surface failure. Parameter estimate information is used to adjust the inner loop's control gains and a command derived from the intercept estimate is fed forward to retrim the aircraft automatically. The tracking performance of this automatic retrimming method which uses system identification, surpasses the conventional approach of exclusively relying on integral action for automatic retrimming. The novel adaptive and reconfigurable flight control system successfully handles a horizontal tail control surface loss.

Section: Introductionmentioning

confidence: 98%

Adaptive and reconfigurable flight control using feed-forward action

Nelson¹,

2006 American Control Conference

2006

“…A certainty equivalence based control law using estimates of the intercept and slope coefficients is proposed. In reference [9], batch estimation, linear regression, and two new methods that incorporate an intercept to account for unmodelled bias are used to estimate the frequency parameter of a sinusoidal waveform. Reference [10] extends the intercept theory developed in reference [9] into a nonlinear regression application.…”

Section: Introductionmentioning

confidence: 99%

Reconfigurable flight control

Proceedings of the Institution of Mechanical Engineers, Part G:

Nelson

2005

Indirect adaptive control for reconfigurable flight control is advocated. Specifically, online static system identification using a moving window/batch estimation is implemented. The ensuing linear regression is augmented with an intercept. The intercept parameter is included to address the effects of trim change associated with the occurrence of a control surface failure. Parameter estimate information is used to adjust the inner loop's control gains, and a command derived from the intercept estimate is fed forward to automatically retrim the aircraft. The tracking performance of this automatic retrimming method, which relies on system identification, surpasses the conventional approach of exclusively relying on integral action for retrimming. These adaptive and reconfigurable flight control concepts are illustrated in the context of a fighter aircraft's pitch plane dynamics. The novel adaptive and reconfigurable flight control system successfully handles a 50 per cent horizontal tail control surface loss.

“…We now augment the parameter vector and include an intercept into the nonlinear regression [4]. After linearization, the augmented formulation is…”

Section: Nonlinear Regression With Interceptmentioning

confidence: 99%

“…This differs from [3], where on-line identification using linear regression is investigated for reconfigurable flight control using an intercept parameter that is introduced to capture offset/change in trim effects caused by failure. Finally, in previous work [4], two new methods that incorporate an intercept to account for unmodelled bias, are used to estimate the frequency of a sinusoidal waveform and its DC component. This paper extends the linear estimation theory developed in [3] and [4] into nonlinear regression.…”

Section: Introductionmentioning

confidence: 99%

“…Finally, in previous work [4], two new methods that incorporate an intercept to account for unmodelled bias, are used to estimate the frequency of a sinusoidal waveform and its DC component. This paper extends the linear estimation theory developed in [3] and [4] into nonlinear regression. The current paper is an attack on nonlinearity by including an intercept parameter in nonlinear regression that accounts for linearization induced truncation error.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Projectile launch point estimation from radar measurements

Nelson

Proceedings of the 2005, American Control Conference, 2005.

Musick

Nonlinear regression with an intercept is investigated and a new nonlinear regression algorithm is developed. The application area considered is ballistic trajectory determination from battlefield radar measurements. Specifically, the geo-location of an enemy artillery piece is pursued. Careful modelling of the nonlinear measurement situation at hand and the inclusion of an intercept parameter in the nonlinear regression shows a considerable improvement over conventional iterative least squares estimation when nonlinearity is dominant. Moreover, the estimation performance does not degrade from standard iterative least squares (ILS) in cases where the nonlinearity is weak compared to the measurement noise in the equation error, provided that the data record is sufficiently long.