{"title":"An Empirical Knowledge Representation Model Based on Similarity Measures of Multi-compression-Layers for Lre Operational State Identification","authors":"Fudong Li, Zijun Liu, Jinglong Chen","doi":"10.1109/ISSSR58837.2023.00024","DOIUrl":null,"url":null,"abstract":"Data-driven intelligent fault diagnosis methods for liquid rocket engine (LRE) operational condition identification have good prospects for application. However, the small number of LRE samples and the highly unbalanced categories pose difficulties for the training of intelligent diagnostic models. To address the problem of the unbalanced distribution of LRE sample categories, an empirical knowledge reconstruction expression model is proposed. Instead of the traditional sample-label one-to-one training model, the model represents the operational status of the input samples by measuring the similarity between the input samples and each sample in the historical knowledge base, combined with the expert knowledge in the historical experience base. The model combines the powerful data mining capability of the neural network model with the precise point analysis capability of the expert knowledge, while the continuously improved historical sample distribution and expert knowledge and experience base provide the scientific basis for the state discrimination and improvement direction of LRE. Further, the validity and engineering value of the proposed method was verified using real LRE test drive data.","PeriodicalId":185173,"journal":{"name":"2023 9th International Symposium on System Security, Safety, and Reliability (ISSSR)","volume":"33 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2023-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"2023 9th International Symposium on System Security, Safety, and Reliability (ISSSR)","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/ISSSR58837.2023.00024","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 0
Abstract
Data-driven intelligent fault diagnosis methods for liquid rocket engine (LRE) operational condition identification have good prospects for application. However, the small number of LRE samples and the highly unbalanced categories pose difficulties for the training of intelligent diagnostic models. To address the problem of the unbalanced distribution of LRE sample categories, an empirical knowledge reconstruction expression model is proposed. Instead of the traditional sample-label one-to-one training model, the model represents the operational status of the input samples by measuring the similarity between the input samples and each sample in the historical knowledge base, combined with the expert knowledge in the historical experience base. The model combines the powerful data mining capability of the neural network model with the precise point analysis capability of the expert knowledge, while the continuously improved historical sample distribution and expert knowledge and experience base provide the scientific basis for the state discrimination and improvement direction of LRE. Further, the validity and engineering value of the proposed method was verified using real LRE test drive data.