An Experimental Study on the Characteristics of Damage Assessment By Neural Networks
When a structure is damaged, its dynamic characteristics are found to be changed. The nervous damage-assesment method enables us to measure the dynamic characteristics from the response signals and use them to assess the degree of structural damage. Although there have been some studies on a certain typical cases, it is rare to find a study about the characteristics of the nervous damage-assesment method or about its applicability or its strength and weakness. So this study researches on the characteristics of nervous damage assesment method and also on its limit in application using various dynamic responses. First, this method shows the degree of damage normally using the rate of profile decreasing damage. Second, what the nervous damage assessment method usually does is to identify the damage of the structure from the signals of the nervous system which has already learned the type of signals from the damaged structure with the same load. However, the loads inflicted on the structures are not always the same. So this study experiments whether the nervous damage-assesment method which is trained with the only one kind of load is able to identify the damage when a different load is inflicted.