Comments on the letter ´Practical problems in the use of correlation analysis to indentify nonlinear systems´

I welcome the invitation of the Editors of IEE Proceedings Part D to reply to the letter by Boland and Doyle. The letter illustrates that a decomposition procedure given by Billings and Fakhouri [2] as part of an identification algorithm for the system illustrated in Fig. 1 cannot be used to provide individual discrete-time estimates of the linear subsystems of a continuous process. While this is a useful observation which arises whenever there is no sampler separating the two linear blocks, it does not invalidate the algorithm in any way and can easily be avoided. However, the statement by Boland and Doyle that the algorithm is only valid if ´the identification is performed in the same time domain as that on which the system operates´ is incorrect, as shown below.