Industrial Big Data Research And Development For Digital Workshops

With the development of data acquisition technology and transmission technology,the amount of data collected by various CNC equipment in the digital workshop is growing at an alarming rate.In recent years,the booming big data technology has encountered the budding intelligent manufacturing,and raised the problems faced by industrial big data at this stage: how industrial big data provides services for intelligent manufacturing,and how traditional big data technology is applied to industrial big data.This article explores these two issues.Firstly,the paper analyzes the service mode of industrial big data,and then focuses on the problems faced by the troubleshooting and maintenance of numerical control equipment and component faults in digital workshop production.The application of gradient boosting algorithm and deep learning algorithm in the numerical control of key components of RUL is emphasized..The main contents of this article are as follows:(1)Service mode of industrial big data and development of microservice interfaceBy analyzing the sources and characteristics of industrial big data,the problems faced at the current stage,and the relationship between industrial big data and intelligent manufacturing,according to the production characteristics of digital workshops,combined with the network physical system,a modular industrial big data service model is designed.It includes four modules: acquisition,calculation,service and application.Combined with the traditional micro-service of the Internet,the micro-service interface of industrial big data is designed.(2)Research on RUL prediction method based on gradient boosting algorithmTaking the gradient boosting algorithm as an example,the tool and bearing RUL prediction model for digital workshop is designed.The whole process is divided into two stages: acquisition and analysis and prediction.The theoretical application and analysis process of big data analysis algorithm in equipment acquisition data preprocessing,feature extraction,feature selection and model training are described.The noise reduction method of wavelet transform and the time-frequency domain of Fourier transform and wavelet packet transform are used.Analytical method.Finally,the feasibility and correctness of the GBL and XGBoost gradient boosting algorithms in the digital component RUL prediction are verified by experiments.(3)Research on RUL prediction method based on deep learning algorithmBy analyzing the impact of input timing on RUL prediction,the deep learning algorithms RNN,LSTM,and GRU are introduced,compared with other supervised learning algorithms,and how RNN takes time series as input,and how LSTM and GRU improve RNN The gradient disappears.Then,a multi-pair GRU neural network model for RUL prediction of tools and bearings was designed,and the effect of GRU on RUL prediction was verified by experiments.