In system identification, the way one excites the system is crucial. For the most direct or indirect system identification methods, a random input signal is used. The randomness, in this case white gaussian and zero mean, should make sure that all unk...
In system identification, the way one excites the system is crucial. For the most direct or indirect system identification methods, a random input signal is used. The randomness, in this case white gaussian and zero mean, should make sure that all unknown system modes are excited. The purpose of this work is to outline a input design such that the identification results are improved. In this paper, a new input design is proposed based on input/output data gathered from gathered from random excitation. The set of data is then used to compose a new set of input data, from which the system is excited and identified. The results obtained from numercal simulations show that the noise sensitivity is drastically reduced and accurate models can be identified, even with high noise disturbances.