Abstract: Aiming at the validation problem of multi-sensor target recognition models，the problem of validating target recognition models using multi-neural-networks by the layered and orderly way is presented，which uses the ability of self-organization and self-learning of neural networks.A neural-network-based method is proposed that first learns key properties of the behaviour of alternative target recognition simulation models，then classifies real system behaviour as coming from one of the models，and gives the credibility evaluation to the models.Simulation results show the feasibility and validity of this method.

Key words: neural networks, target recognition, model validation

摘要： 针对多传感器目标识别仿真模型的验证问题，提出了一种基于多神经网络的“分层有序”的模型验证方法。该方法利用神经网络的自组织和自学习能力，通过对各种目标识别模型关键行为特性的学习，将实际系统行为归类为其中的一种模型，从而对模型的可信性做出评估。仿真结果进一步说明了该方法的可行性和有效性。

关键词: 神经网络, 目标识别, 模型验证