Research on fault diagnosis technology of simulated altitude test facility based on multi-optimization strategy, real-time data transfer, and the M-H attention-RF algorithm

作     者：Zhou, Qifan Guo, Yingqing Zhao, Wanli Xu, Kejie Wang, Kun Wu, Zhenglong Sun, Hao 

作者机构：Northwestern Polytech Univ Xian 710000 Peoples R China Xian Inst Modern Control Technol Xian 710054 Peoples R China 

出 版 物：《MULTIMEDIA TOOLS AND APPLICATIONS》 (多媒体工具和应用)

年 卷 期：2024年第83卷第10期

页      面：28729-28760页

核心收录：

学科分类：0808[工学-电气工程] 08[工学] 0835[工学-软件工程] 0812[工学-计算机科学与技术（可授工学、理学学位）] 

基　　金：National Science and Technology Major Project, (J2019-V-0003-0094) National Science and Technology Major Project 

主　　题：Simulated altitude test facility Data extraction and transfer techniques Multi-optimization strategies Granularity improvement schemes Fault diagnosis algorithms 

摘      要：The simulated altitude test facility, as an important means to verify the performance, characteristics, and evaluation result criteria of aero-engines, has a pivotal engineering significance in the process of aero-engine development and promotion of application. In order to cope with the drawbacks of traditional techniques for experimental processes, this paper proposes the real-time data extraction and transfer techniques with multiple optimization strategies and the fault diagnosis technology of simulated altitude test facility with an improved optimization algorithm is propose, Firstly, the optimization strategy based on peak shaving + peak fast processing and token bucket instructions with multi-threaded parallel processing flow allocation call logic is used to realize the test data for fast extraction and migration demand, and then the overall data transfer function is optimized in granularity improvement schemes by using the abstraction optimization strategy mechanism based on Direct Routing mode to maximise real-time targets while ensuring correspondence and completeness of test data. Finally, the random forest algorithm with Multi-Head Attention optimization is used to implement the diagnostic technology research of the simulated altitude test facility under two scenarios under the data-driven mode, and the analytical comparison and validation results with the unimproved and optimized Random Forest algorithm are given. The results indicate that the amount of test data synchronization reaches 300 + lines per second, the accuracy of fault diagnosis identification is increased by 30% at the highest degree, and the proposed improvement research has a very high degree of application value and innovativeness.