Research On Demand Forecasting Method For Spare Engine Of Civil Aviation Driven By Condition Data

The civil aero-engine fleet is generally composed of multiple engines of the same model.To meet the fleet's scheduling,operation and maintenance needs,airlines must be equipped with a certain number of spare engines.The number of spare engines directly affects the operating cost and guarantee rate of the fleet.Excessively high or too low numbers of spare engines may cause huge cost losses or waste of resources.With the continuous expansion of the fleet size,airlines' demand for a spare engine demand forecasting system is becoming stronger,so it is of great practical significance to study the method of determining the number of spare engines suitable for the civil aviation engine fleet.The process of determining the number of spare engines is affected and constrained by many factors such as the health of the fleet,engine disassembly plans,scheduling plans,etc.And because most of these factors are closely related to the engine condition data,this paper conducts research on the demand forecasting method for spare engine of civil aviation driven by condition data.This paper proposes a method for predicting remaining useful life of engine based on deep convolutional neural network(DCNN),Since convolutional neural network is very suitable for processing variable and complex signals,it is used to mine hidden features of engine condition data.A novel network model structure is designed according to the characteristics of the state data,and a piecewise linear regression model is used to set the remaining life label for the training set samples.Aiming at the difference in operating cycles of different engines,the data is reconstructed using the variational autoencoder(VAE),an unsupervised reconstruction method is used to identify the abnormal initial position of the data sequence,and the life tag platform values are corrected.Use sliding windows to prepare data samples and input them into the trained network model,then output the remaining life prediction results.The C-MAPSS simulation data set is used to verify that the proposed method has high prediction accuracy.This paper decomposes the work of making engine disassembly plan into two parts: the forecast of dismantling period and the decision of the scope of maintenance work.Use real engine performance decline data samples to prepare a performance decline case library,and estimate the life time node by analyzing the time series decline trend,then comprehensively consider the prediction results based on performance degradation and the results based on life limited parts(LLP)replacement rules,airworthiness directive(AD)constraints,service bulletin(SB)constraints,and hardware damage of critical parts,to obtain a knowledge and data fusion driven dismantlement period prediction model.Based on the knowledge model of unit maintenance level,a fuzzy evaluation method of maintenance work range based on state parameters is proposed,and the mapping relationship betwe en state parameters and available maintenance levels is established to determine the maintenance work scope of the engine.This paper proposes a forecasting method for the fleet's spare engine demand based on improved ant colony algorithm for the individual needs of airlines,many factors such as the health status of the fleet,replacement plans,and scheduling plans are incorporated into the process of determining the number of spare engines.Establish a fleet total support cost evaluation model,and optim ize the colony algorithm's pheromone update mechanism,so that it can search for the optimal solution with high support rate and low cost in the solution space.Comparison with GE's traditional methods proves that the method proposed in this paper can effectively reduce operation costs under the condition that the guarantee rate meets the requirements of operation and maintenance,and can guide the actual operation and maintenance work such as adjustment of scheduling schemes and determination of the number of spare engines.On the basis of completing the above theoretical research,with "customizable aero-engine health management and maintenance decision support system" as a basic platform,according to the actual needs of airlines,design and develop civil aviation engine fleet spare engine demand prediction system components to provide technical support for airlines to carry out fleet remaining life prediction,disassembly plan formulation,fleet spare engine demand prediction and other fleet scheduling management work.The research in this paper is of theoretical guiding significance for enriching the reserve engine demand forecasting methods of civil aviation engine fleet s,improving the accuracy of remaining useful life prediction,optimizing fleet scheduling schemes,reducing fleet operation and maintenance costs,and improving fleet reserve engine guarantee rate.