Introducing a Combined Equation/Output Error Approach in Parameter Estimation

Abstract

This investigation presents a new approach in the field of parameter estimation. The well known equation and output error estimation methods are combined in a flexible manner into one estimation approach where the model response and measurement comprise both state and non-state parameters of the system under test. Various simulation tests are conducted based on linear, non-linear, flight mechanically unstable aerodynamics with and without noisy signals. The new approach shows a well balanced matching performance both in the equation and output error domains. Furthermore, it is more stable in the iteration process than the output error method due to the equation error ingredients. The estimated parameters of the pure output error approach converge most closely towards the true ones but this approach is most sensible and prone to non-convergent solutions. In contrast to this, the combined equation/output error approach gives flexibility in the estimation process where by means of a suitable weighting among the state and non-state observation vector elements any estimator behavior can be emulated from a pure equation error to a pure output error approach.