运载火箭测试数据分析与故障诊断方法

Data Analysis and Fault Diagnosis Methods for Launch Vehicles

摘要:

针对如何从运载火箭大量历史测试发射数据中发掘有用信息的问题，提出了一种基于数据挖掘的运载火箭数据分析与故障诊断方法，为火箭的故障诊断、产品设计、状态检测提供服务。针对火箭数据的特性和实际业务分析的需求，使用基于皮尔逊系数的相关性分析方法、基于希尔伯特变换的包络分析方法、基于窗口滑动函数的故障诊断方法组建火箭数据分析平台，对运载火箭的数据进行了深层次的挖掘和诊断。采用某型号火箭测试数据对火箭数据分析平台进行了验证，结果表明:以数据驱动的火箭数据分析平台相关性分析准确、参数包络线绘制精准、可有效识别异常数据，分析结果与理论知识相符，具有较高的实用价值，相比于传统数据分析方法更为精准、全面。

Abstract:

A data mining based data analysis and fault diagnosis method for launch vehicles is proposed to explore useful information from a large amount of historical test launch data, providing services for rocket fault diagnosis, product design, and status detection. In view of the characteristics of rocket data and the needs of actual business analysis, the rocket data analysis platform is established by using the correlation analysis method based on Pearson coefficient, the envelope analysis method based on Hilbert transform, and the fault diagnosis method based on window sliding function. The rocket data analysis platform is verified with the test data of a certain type of rocket. The results show that the correlation analysis of the data driven rocket data analysis platform is accurate, the parameter envelope is accurately drawn, and the abnormal data can be effectively identified. The analysis results are consistent with the theoretical knowledge, which has high practical value. Compared to traditional data analysis methods, it is more accurate and comprehensive.