Accuracy Evaluation of Least-Squares Method for a Class of Systems with Multi-Errors in Input and Output*

This paper aims to evaluate the identification error of Least Square (LS) method for a class of systems with bias error, scale error and random noise in the measurement of both control input and output. According to the relation between the LS estimation error and the sensor deviations, this paper proposes an accuracy evaluation method to provide the bound of the estimation error for LS. Besides, the robustness of the evaluation result to the prior information is analysed. The theoretical analysis is verified by the simulation on the linearized longitudinal aircraft aerodynamic model.