基于机器学习的固体火箭发动机无损检测

传统固体火箭发动机无损检测图像判读工作存在人工识别效率低、图像数据分散及数据利用率低等问题。本文借助机器学习算法与计算机视觉技术，利用大量发动机无损检测图像数据开展无损检测图像数据预处理、边缘检测以及数据模型训练和应用等技术研究，探索快速、准确获得发动机无损检测图像数据特征的方法，深入挖掘固体发动机无损检测数据的内在联系，找到潜在规律。本研究不仅为固体发动机无损检测图像判读提供了一种准确、高效的手段，同时，能够为发动机无损检测图像识别、测量、判读和发动机相关故障模式分析与故障诊断提供数据和决策支持，也能够为未来机器学习在固体发动机无损检测图像判读领域的深入应用提供实践探索和理论研究方面的参考。

Intelligent interpretation of nondestructive testing data of Solid Rocket Motors based on Machine Learning

The traditional interpretation of images of nondestructive tests of solid rocket motors encounters such problems as poor efficiency of manual identification, scattered image data, and low utilization rates of the data. This study processed data from a large number of images of nondestructive testing, detected the edges, trained a data model, and applied it using a Machine Learning algorithm and computer vision technology. The aim of this study is to explore fast and accurate methods to obtain data characteristics of nondestructive testing of such motors, mine the internal relations and potential laws of such data, and provide data and decision support to interpret images of the nondestructive testing of such motors and inform fault diagnoses. The results provide a reference for the application of Machine Learning to the interpretation of images of the nondestructive testing of Solid Rocket Motors.