Development of preprocessing system for non-stationary reactor noise data recording

This report is concerned with the signal preprocessing techniques used in an on-line data recording system developed for acquisition of non-stationary reactor noise signals. The reactor noise signals recording requires signal preprocessings such as (1) separating the signal into fluctuation and dc-component and (2) amplifying the fluctuation with an optimal gain. However, conventional signal preprocessors using analog devices could not work well for the non-stationary signals. By applying a digital signal processing technique, the author has developed a new signal separator which is capable of changing the type and characteristics of filters depending on the signal properties. A filter-amplifier with function of searching for the optimal gain has been also developed. Both equipments were tested through data recording of simulated and real reactor noise signals. The test results show that these equipments make it possible to record the noise signals not only in the case where the dc-component slowly changes but also in the case where it changes rapidly with a large amplitude. (author)