Design And Implementation Of An Auxiliary Decision-making System For Aero-Engine Testing

The testing technology for aero-engines is a key factor in advancing aero-engine development.With the development of aero-engine industry in China,aero-engine testing data presents characteristics of large volume,multi-source,and heterogeneity.The existing testing data management system lacks unified management of multiple-source heterogeneous data and real-time data processing and correlation analysis capabilities.Technical personnel find it difficult to make judgments on the real-time data of the testing site and lack sufficient analysis capabilities for testing result data.In view of the above problems,this thesis proposes an auxiliary decision-making system for aero-engine testing as a solution,providing correlation analysis,testing status analysis,and decisionmaking support capabilities to provide scientific methods and basis for testing site decision-making and improve the analysis efficiency of testing result data.The main work of this thesis is as follows.1.To address the needs of correlation analysis and anomaly detection for testing data,this thesis has established multiple auxiliary decision analysis models.Firstly,the engine operation status recognition model is established to identify states such as zero-point,cold running,and start-up.Secondly,the threshold alarm model is used to provide warnings for sensors that exceed the limit range under different testing operation states.Thirdly,the rule filtering model utilizes Python expressions to express parameter relationships at multiple levels such as testing machine and equipment operation principles and testing methods.Expert knowledge is then used to sort out rules for parameter correlation relationships,parameter variation relationships,and physical constraint relationships such as total pressure,total temperature,flow rate,and thrust.These rules are then used for multi-parameter anomaly detection.Fourthly,an unsupervised time-series anomaly detection model is designed to analyze the curve characteristics of sensor anomaly data and implement an algorithm for detecting abnormal intervals of key sensor data.Finally,based on testing fault data,various time-series correlation analysis algorithms are compared,and the best performing algorithm is selected to recommend anomalous correlation sensors.2.To address the issue of data silos in testing subsystems,this thesis has utilized ETL technology and Doris database to establish an efficient data integration platform.The platform integrates multiple heterogeneous data sources into a standardized data model and stores them in a unified data warehouse.Furthermore,the testing data is cleaned,deduplicated,and verified to ensure data quality and consistency.The platform provides fast,accurate,and reliable data query services and serves as the foundation for data-driven auxiliary decision analysis.3.To address the issue of insufficient testing data analysis capabilities,this thesis has designed and implemented algorithm modularization methods and visualization modeling methods,and provided various forms of visual display of analysis results.These methods aim to reduce the threshold for testing personnel to develop and use data analysis models,enable the system to increase the number of analysis models in the future.As a result,the utilization capability of aero-engine testing data has been greatly enhanced.Based on the work,in combination with practical business scenarios,this thesis conducted a requirements analysis,overall design,detailed design,and development implementation of the auxiliary decision-making system for aero-engine testing.Finally,functional and performance testing of the system was carried out,demonstrating its ability to meet the real-time data analysis and anomaly detection needs of aero-engine testings,and providing testing personnel with decision-making support information.