Research On Online Warning Technologies Of Tool Wear Driven By Big Data

In drilling process of CFRP/ titanium alloy laminated materials,different cutting performance of materials resulting in serious tool wear and make it difficult to guarantee production efficiency.Traditional methods mainly depend on the operators.Affected by operators' experience,the reliability is low and not suitable for automatic processing.With a large number of automatic equipment has been utilized in the workshop,it is urgent to monitor tool wear condition in the manufacturing process so as to provide a basic reference for tool change in production.However,during the process of analyzing the massive monitoring signals,the real-time storage and efficient processing are confronted with severe challenges.In-depth research on these aspects is needed so that big data could play an important role in online warning of tool wear.This paper analyzes the status of tool wear during hole making process of CFRP/ titanium alloy laminated material.An online tool wear monitoring scheme based on acoustic emission signal is proposed.In Addition,the data volumn of AE siginal is large when monitoring many facilities simultaneously,it is obvious that big data technologies should be adopted to meet such requirements.On this basis,the platform for signal analysis based on Hadoop were established.A novel online early-warning mode of tool wear driven by big data was put forward.The wavelet packet transform was used to filter the original AE signal and extract the AE characteristics which closely related to tool wear.The selected features were used as the input vectors of the BP neural network and the tool wear model is established.Due to the disadvantages of BP neural network,ant colony algorithm was used to optimize BP neural network.The distributed storage of original signal files relied on HDFS,and the signal feature extraction and modeling algorithms were implemented by MapReduce.On this basis,big data integration framework of signal acquisition,distributed storage and parallel analysis was conducted.The development of online tool wear warning prototype system was completed,and its integration with the MES system was given.The results show that the tool wear model based on neural network has high accuracy in the classification and prediction of tool wear.The online warning mode of tool wear driven by big data satisfies the requirement of monitoring of multiple tools at the same time,accelerated the speed of signal analysis.MES system integrated with online tool wear warning system significantly improves the process monitoring ability.