孪生神经网络在航天产品电性能测试方面的应用

Application of Siamese neural network in electrical performance test of aerospace products

航天产品电性能测试是在外加激励的情况下,测试航天产品不同部位的信号反馈情况,从而确定各部件的质量情况,是检验航天产品质量的重要手段之一。但是,测试得到的信号通道多、时序长、十分复杂,依赖传统信号处理方法难度较高,目前处理方式采用人工统计,效率低、易出错,亟需借助深度神经网络的方法来提高效率和准确率。本文将孪生神经网络引入航天产品电性能测试领域,将数据信息解析为图像信息,转化为图像识别问题;对几种主干网络模型进行对比分析,最终选择ResNet50作为主干网络。实验结果表明,在相同实验条件下,研究的算法能够有效地识别78个通道信号的波形信息,识别准确率达到90.385%,提高航天产品复杂测试信号的识别能力以及智能化水平。

The electrical performance test of aerospace products is to test the signal feedback of different parts of aerospace products with external excitation,so as to determine the quality of each component.It is one of the important means to test the quality of aerospace products.However,the signal obtained from the test has the characteristics of many channels,long timing and complexity,and it is difficult to rely on traditional methods to process the signal.At present,the processing method adopts manual statistics,which is inefficient and error prone.It is urgent to improve the processing efficiency and accuracy with the help of deep neural network.In this paper,the Siamese neural network is introduced in electrical performance test of aerospace products,and the data information is resolved into image information and transformed into image recognition problem.Several backbone network models are compared and analyzed,and ResNet50 is finally selected as the backbone network.The experimental results show that under the same experimental conditions,the algorithm proposed can effectively identify the waveform information of 78 channel signals and the recognition accuracy reaches 90.385%,improving the recognition ability and intelligent level of complex test signals of aerospace products.