Real-Time Data Acquisition And Abnormal Point Analysis Of Rolling Mill

In the rolling mill,the rolling mill is in a state of high load and high speed for a long time,so it is easy to break down.In addition,the rolling mill equipment is composed of complex parts and components,and its volume is huge.If the rolling mill breaks down,it needs complex maintenance work,which will lead to the suspension of production and cause great losses.Therefore,by collecting key data information in real time and carrying out anomaly detection,accidents can be effectively prevented and an important basis can be provided for finding the cause of the failure.The main work of this paper is as follows:Build a communication platform with Siemens S7 series PLC,capture and analyze the messages communicated with Siemens S7 series PLC.According to the parsed message,a data acquisition algorithm is designed to quickly collect custom data.Aiming at the situation of long data acquisition time and large data storage in production site,a data compression algorithm is designed by using Huffman coding.The two algorithms are tested,and the results show that the acquisition algorithm designed in this paper can basically meet the collection of rolling mill production data,and the designed datacompression algorithm can compress rolling mill data well.The abnormal points of the collected rolling mill data are analyzed in the form of curves which are summarized as the combination of sudden abnormality,trend abnormality and frequency abno,rmality.The detection algorithms for these three types of anomalies are studied,and the performance of these algorithms in detecting anomalies in historical data and real-time data is analyzed respectively.According to the performance of these detection algorithms,historical data anomaly detection algorithm and real-time data anomaly detection algorithm are designed respectively.For the historical data detection algorithm,the improved LOF algorithm,the optimal segmentation method and the extreme point method based on filtering are mainly used.For the real-time data anomaly detection algorithm,the differential box diagram method based on sliding window,the optimal segmentation method based on sliding window and the quartile aggregation method based on sliding window are mainly used.These two anomaly algorithms are used to test the data of rolling mill collected in the field.The results show that these two algorithms can detect anomalies well.In the MFC development environment,the data acquisition and analysis software of rolling mill is studied and designed.Through the running test of the software,the results show that the software can collect data from any Siemens S7 series PLC.At the same time,the software can detect the data in the rolling process online/offline.