Research On The Health Monitoring And Life Prediction Technology Of Smart Electromechanical Equipment Oriented To Big Data

With the development of science and technology,the electromechanical equipment systems in the fields of aerospace,manufacturing,energy,metallurgy and other fields are becoming more intelligent and complicated.Traditional maintenance strategies have problems such as "under-maintenance" or "over-maintenance",which are difficult to meet actual maintenance.demand.In order to ensure the safe and reliable operation of the equipment system,under the background of the current big data era,it is of great significance to study the data-driven Remaining Useful Life(RUL)method of the electromechanical equipment system for the application scenarios of the electromechanical equipment system under complex conditions.Therefore,in the context of big data and combined with deep learning theory,this paper proposes a RUL prediction method for electromechanical equipment based on multivariate CNN and a RUL prediction method for electromechanical equipment based on multivariate CNN-LSTM,which solves the problems of traditional methods to a certain extent.Relying on the experience of domain experts,the inability to obtain deep features,and the lack of versatility,etc.,through comparative testing on the PHM2012 data set,better results have been achieved.The main contents of the paper are as follows:(1)Study the degradation characteristic index of electromechanical equipment and construct the input characteristic vector.With the help of the full life cycle vibration signal data of 17 sets of bearings under different working conditions in the "IEEE PHM 2012" data set,a number of typical time domain characteristic parameters are compared and analyzed from the horizontal and vertical directions,and finally the root mean square,standard deviation,The crest factor and kurtosis are both used as bearing degradation indicators.The5-fold cross-validation method is used to divide the data set,and the single degradation feature index is spatially reconstructed,and the construction result is used as the input feature vector of the model.(2)Study the method of predicting the remaining service life of electromechanical equipment based on the multivariate CNN model.By summarizing and analyzing the research of many scholars on the CNN model,drawing on the advantages of their proposed CNN model to build a multivariable CNN model,and comparing and comparing the two indicators with Mean-DNN,RFR,SVR and LASSO through MAE and RMSE.The analysis shows the superiority and feasibility of the prediction model based on multivariate CNN.(3)Research on the method of predicting the remaining service life of electromechanical equipment based on the multivariate CNN-LSTM model.By summarizing the advantages of the CNN model and the LSTM model,a method for predicting the remaining life of electromechanical equipment based on the multivariate CNN-LSTM model is proposed.The constructed feature data matrix is processed as the input of the CNN model,and the CNN model extracts deep features.,Input into the LSTM network to predict the remaining service life,and measure the prediction performance of the CNN model and the CNN-LSTM model through the two indicators of MAE and RMSE,and finally illustrate the feasibility and superiority of the multivariate CNN-LSTM prediction model.(4)Design and implement a big data-oriented electromechanical equipment health monitoring system.Based on the multi-variable CNN-LSTM prediction model,the algorithm function module and Hadoop platform are built,combined with the actual situation to analyze the business process in the system,the back-end is written in Python language,the front-end page is written in lay UI,the health monitoring system of electromechanical equipment is designed and implemented,and Through the PHM2012 data set,verify the core functional modules of the system.