| As a representative of a country’s comprehensive national strength,aero-engine is the core technology to ensure national defence security and strategic advantages.Given that the fuel distributor is one of the critical components in the aero-engine fuel system,higher requirements have been put forward for its reliability,safety,and economy in operation.Due to the lack of monitoring methods for fuel distributors,there are few research achievements in the health management of fuel distributors in China,which brings great difficulties to the maintenance work of fuel distributors.In this study,based on the construction of the fuel distributor performance test bench,research on health status assessment and life prediction of fuel distributors was carried out to determine the weights for each indicator and the corresponding health threshold,improve the cloud model structure,and optimize the cloud focus evaluation method.Based on the evaluation results,the judgment was made on whether to predict the lifespan of the equipment.At the same time,the life prediction model based on the Deep Belief Network(DBN),Whale Optimization Algorithm(WOA),and Bidirectional Long Short-Term Memory(Bi LSTM)was designed to scientifically assess the health status and predict the life of fuel distributors.Firstly,the overall plan for evaluating the health status and predicting the lifespan of the fuel distributor was designed,and the performance test bench for the fuel distributor was built.By operating the test bench,the operational status of the fuel distributor was simulated,and the Lab VIEW CNC acquisition system was used to monitor the operating status of the fuel distributor and save its operational data.Secondly,based on the complex indicators of fuel distributors,the improved cloud model for evaluating the health status of fuel distributors was proposed to represent the ambiguity between adjacent state levels better and effectively improve the accuracy of model evaluation by improving the cloud model structure.In order to solve the problem of subjective experience affecting the threshold of the cloud-based interval,the combined weighting method was used to combine the weights of the subjective and objective indicators of the entropy weight method and the three-demarcation method.The K-means algorithm was used to cluster the measured data to obtain the threshold of each indicator state interval and determine the cloud-based interval.Through example verification,it was concluded that this method is reasonable and effective,which can provide convenience for maintenance personnel to understand the operating status of the fuel distributor.Thirdly,in response to the low accuracy of a single life prediction model,DBN was adopted to extract equipment degradation features to construct health indicators,WOA was used to optimize Bi LSTM network parameters and health indicators were input into the optimized model for life prediction to construct a DBN-WOA-Bi LSTM life prediction model applied to fuel distributors.It was concluded that the prediction accuracy of this model is higher,and it is more consistent with the performance degradation trend of the fuel distributor,so it can accurately predict the remaining life of the fuel distributor.Finally,a simulation platform for evaluating the health status and predicting the lifespan of fuel distributors was developed.Using QT5.12 as the development environment for the software system,the My SQL database stores the inspection data of the fuel distributor,and the simulation software was selected to make maintenance decisions.It was concluded that the software system is straightforward to understand,and it can obtain health status assessment and life prediction results based on interface guidance,making it easy for maintenance personnel to operate.The feasibility of the method proposed in this study was confirmed during the use of the system,which played an important role in evaluating the health status and predicting the lifespan of fuel distributors. |
PDF Full Text Request