Identification of submerged vehicle dynamics through a generalized least squares method

This paper discuses the application of a generalized leaset-squares method for post-flighttest analysis of small submerged craft. Specifically, the purpose is to identify the transfer functions characterizing the longitudinal and lateral dynamics from exprimental data pertaining to appropriate maneuvers. The method utilizes a set of signals, called measurement signals, generated by processing the input-output variables by a cascade of know digital filters. An enginvector of the Gram matrix of the measurement signales yields, after a linear transformation, the estimated coefficients of the desired transfer function. The method offers the advantages of global minima and robustness of estimates relative to data noise.