基于动量BP神经网络激光陀螺调腔检测方法

Detection Method of Laser Gyroscope Cavity Adjustment Based on Momentum BP Neural Network

为解决激光陀螺人工调腔质量低、效率不高等缺点,提出一种由CCD相机和光电倍增管构成的多传感器信息融合调腔检测方法,该方法同时检测光斑、光阑中心点及陀螺损耗值,并将这些信息送入融合中心,经过融合计算得到陀螺调腔是否合格的综合判断。融合系统采用动量BP神经网络对多源信息进行融合,根据调腔过程中的输入、输出情况,设计包含输入层、隐含层及输出层的3层网络拓扑结构。实验结果表明,此种方法对激光陀螺调腔质量是否合格判断准确率为93.81%,比人工采用单一传感器分步检测准确率提高了6%。

In order to solve the manual detection drawbacks of laser gyroscope cavity adjustment,such as low quality and low efficiency,a multi-sensor information fusion detection method is proposed using a CCD camera and a photomultiplier.The center of the facula and the diaphragm and the loss of laser gyroscope are obtained and then transmitted to the fusion center.After fusion calculation,the integrated judgment is produced.The fusion system utilizes the momentum back-propagation neural network（BPNN） to fuse the multi-source information and output the final decision.And according to the modes of the detected signals and output decision,a three layers topology structure including an input layer,a hidden layer and an output layer is designed.The experimental results indicate that the accuracy of the proposed cavity adjustment detection method is 93.81%,which is higher than the manual step detection method using a single sensor about 6%.