Research On Trend Analysis Method Of Aircraft Pump Source System State Change

The aircraft pump source system has high working intensity and harsh working environment.It directly affects the flight safety of the aircraft.The trend analysis and prediction of its state change has very important theoretical and engineering application value.In this thesis,a certain aircraft pump source system is taken as the research object,and the monitoring parameter data collected by sensors is fused according to the information fusion technology.The multi-joint entropy is obtained by quantifying the degree of the system disorder to characterize the overall operating state of the system.In order to analyze the state change trend of the pump source system,the ARMA-ELM prediction model is first designed and validated in this thesis.By calculating the mean sequence of each monitoring parameter,it is used as the input of the ARMA-ELM model to predict the multi-joint entropy of the pump source system.The research shows that the designed ARMA-ELM prediction model can make a more accurate prediction of the system state indicators.Then,the trend analysis of the multi-joint entropy data sequence is carried out based on the GRU neural network method.The results show that the designed GRU prediction model can effectively represent the state change trend of the aircraft pump source system.In order to further improve the prediction accuracy,the combination prediction theory is studied in this thesis,and a combined prediction model based on ARMA-GRU is proposed to predict the multi-joint entropy data.The research shows that the designed ARMA-GRU combined prediction model can not only reflect the overall trend of multi-joint entropy data,but also reflect the characteristics of local fluctuations,and the prediction accuracy of aircraft pump source system state index is very high.A comparative study on the trend analysis methods used in this thesis is also conducted.The results show that the accuracy of ARMA-GRU combined prediction model is higher than that of GRU prediction model,and the effect of GRU prediction model is better than ARMA-ELM prediction model.