Research On Hadoop-based Analysis Platform And Analysis Technology Of Hobbing Machine Fault Information

Under the background of the whole industrial system tending to the full intelligence, big data has already become a distinctive feature of manufacturing. The development of IOT technology makes that more and more information can be collected. These data can reflect the real-time change of the health status and performance of the equipment, which contains a large number of fault knowledge. However, due to the industrial big data is too fragmented, redundant, heterogeneous and low-value, it is not effectively used. Traditional fault diagnosis technology has been unable to deal with effectively. How to use big data technology to analyze the fault information of mechanical equipment becomes a difficult problem. Therefore, in order to improve the level of intelligent fault diagnosis, this paper studied mechanical equipment fault information analysis technology based on big data technology.Considering fault information characteristics of hobbing machine, we studied the existing problems in the analyzing fault information of hobbing machine. And then we proposed the overall framework of the platform fault information analysis of hobbing machine. In order to integrate fault information, a unified data environment is established. To solve the problem of multi parameters and large data, fault information is processing concurrently based on MapReduce architecture. Fault feature is fused based on PCA. And parallel random forest model is realized based on MapReduce architecture in order to solve the big dataset that traditional fault diagnosis technology cannot deal with.Based on the above research results and the support of the platform of fault information analysis of hobbing machine, an experiment platform based on Hadoop is established. And the parallel random forest fault classification algorithm based on MapReduce architecture is realized. With five different types of fault data, data preprocessing is based on cosine similarity measure. And the principal component analysis is used to obtain the fault information feature. After getting the parallel random forest model, test set is used to test. The results show that classification rate continues to increase. And the validity of the method is proved.