针对弹载计算机硬件验收过程中,人工分析测试数据耗时长、易误检等问题,提出一种基于PNN网络的弹载计算机故障诊断方法。该方法首先将测试数据由矩阵形式简化为向量,然后通过编码映射,计算出对应的故障编码输出值,最后利用PNN网络对简化后的向量和故障编码输出值进行训练,实现不同测试数据的故障分类。通过对某型弹载计算机试验数据进行仿真,验证了该方法可行性及准确性。

A fault diagnosis method of missile-born computer based on PNN is proposed to avoid time loss and false drop caused by manual testing. Firstly, the method simplifies the test data from matrix form to vector,and gets coded value for the types of falut by code mapping, finally identifies the fault category with the model trained by PNN. Simulation results from the testing data of missile-borne computer prove the feasibility and accuracy of the method.