Signal processor for evaluating test data e.g. in nuclear reactor testing - uses neural network with interconnection weighting factor correction
摘要
An input device receives a measurement signal from a test device and an evaluation device analyses it to detect the occurrence of a fault in and/or a deposit on the tested object. The evaluation device contains a neural network signal processor with a number of processing units, each with a defined input/output characteristic, and a number of connecting devices with individual transfer characteristics These interconnect the units with defined weighting factors. The processing device has a learning function for correcting the weighting factors to optimise the unit output signals. USE/ADVANTAGE - For non-destructive test, e.g. in nuclear reactor steam generator. Automatic data evaluation is achieved as well as increased evaluation accuracy and efficiency.