Abstract: This paper presents an identification approach based onneural network method with sub-regions to identify damages in a rectangularplate using the LM optimized algorithm. The numerical results of simulationsfor the cantilever plate with some damages show that this approach is betterthan that from the conventional method with single network. Since thismethod realizes the synchronous output of position and intensity of damagesin the structure, this approach does not need many sub-networks appeared ina conventional hierarchical identification method.