Optimal Test Inputs for Helicopter System Identification

[摘要] The test input applied to a helicopter, or any other system, for the purpose of system identification can have a substantial effect on the parameter estimates obtained. It is therefore important that an appropriate input is chosen. Inputs must take account of the requirements, and restrictions, of the application. For example, in the rotorcraft case studied a linearised model is being identified, and it is therefore essential that the input produces a linear response. A straightforward method has been developed for the design of multi-step inputs. This method is based in the frequency-domain, and involves tailoring the auto-spectra of the inputs to give long, linear test records, and parameter estimates with reasonably low variances. In flight trials using the Lynx helicopter at RAE (Bedford), the double-doublet input, designed with this method, has been found to be a significant improvement over more traditional inputs. Using the data from the flight trials of the double-doublet, both equation-error and output-error identification have been carried out. Several discrepancies were found between the theoretical and identified models. More work is required to clarify this. Numerical difficulties were encountered during the output-error identification, and these were attributed to ill-conditioning resulting from the use of an unstable system. The design of optimal inputs has also been investigated. In particular, constraints have been developed which are suitable for ensuring that the optimal inputs produce linear responses, and are robust. Conventional energy constraints were found to be of little use for these purposes. Algorithms have been developed for the design of optimal inputs with a variety of constraints, and simulation studies have been made to gain an understanding of the effect of these constraints on the form of the inputs. With the constraints obtained from this work, an optimal input has been designed for use with the Lynx helicopter. This input is robust, and yet is predicted to give significantly improved parameter estimates. Unfortunately, at the time of writing, flight trials of this input could not be performed.

[发布机构] University:University of Glasgow

[关键词] Electrical engineering, Aerospace engineering