装备试验大数据治理体系构建及其对质量评估精度的提升路径研究

The Study on the Construction of Equipment Test Big Data Governance System and Its Path to Improve Quality Evaluation Accuracy

摘要:

该研究针对装备试验中多源数据孤岛、质量评估精度不足的问题，对装备试验质量评估的技术路径进行了研究。通过构建 “采集 - 治理 - 融合 - 应用” 四级装备试验大数据治理体系，分析某型装备发射装置达到标准次数的试验数据，并提出基于可信度加权的多源数据融合方法，结合改进 CNN-LSTM 模型实现装备试验质量评估；经分析发现传统评估方法因数据维度单一、特征提取不充分，缺陷诊断准确率相对较低，实证测试显示新体系下质量评估准确率均能够达到标准，较传统 BP 神经网络提升了相应的百分点，诊断耗时大大缩短。该研究提出的装备试验大数据治理体系与质量评估方法，满足装备试验质量评估的实时性需求，为装备试验质量评估提供了高效可行的技术路径。

Abstract:

This study addresses the issues of multi-source data silos and insufficient accuracy in quality assessment during equipment testing, and conducts research on the technical path for equipment test quality assessment. By constructing a four-level big data governance system for equipment testing, namely the "Acquisition-Governance-Fusion-Application" system, it analyzes the test data of a certain type of equipment launcher when it reaches the standard number of tests, and proposes a multi-source data fusion method based on credibility weighting. Combined with an improved CNN-LSTM model, this study realizes the equipment test quality assessment.The analysis reveals that traditional assessment methods have relatively low defect diagnosis accuracy due to single data dimension and insufficient feature extraction. Empirical tests show that the quality assessment accuracy under the new system can all meet the standards, which is a corresponding percentage point higher than that of the traditional BP neural network, and the diagnosis time is significantly shortened. The big data governance system and quality assessment method for equipment testing proposed in this study meet the real-time requirements of equipment test quality assessment and provide an efficient and feasible technical path for equipment test quality assessment.