Design And Implementation Of Fault Diagnosis System For Vertical Roller Mill

Vertical roller mill is one of the most important equipments in the cementproduction. Its running condition is directly related to its safety and efficiency.Vertical roller mill usually work under complex conditions and harsh environmentswith complex running condition. Generally, a monitoring system of vertical rollermill will monitor a range of signals, including vibration signal, displacement signal,temperature signals. Due to the non-stationary, diversity and synthesis of the causesto the mechanical fault, it is difficult to decide the mapping from a signalcharacteristic to a mechanical fault. Existing fault diagnosis of the vertical roller millmainly relies on the experience and knowledge of experts in related fields. It leads tothe lack of real-time, intelligent and accuracy. This paper combined the technologiesof FBG sensor network, data storage and analysis, pattern recognition and mechanicalcharacteristic of vertical roller mill to achieve the intelligent condition monitoring andfault diagnosis for vertical roller mill. The main work is as follows:(1) Taking the vertical roller mill mechanical characteristics and its actualworking environment into account, the model of vertical roller mill was constructedand the key mechanical characteristic has been analyzed. As the object of study, themill housing has been analyzed using the Finite Element Analysis (FEA) function.The natural frequency and random vibration response was gain from the simulation.Through the load simulation of the rocker，the its distribution of stress-strain was gain.On the basis of the existing experience and knowledge, the main types of fault wereextracted and the relations between the fault types and signal characteristics werepredicted.(2) The natural frequency, stress-strain, threshold table and relation betweensignal characteristics and mechanical fault were used as input of fault diagnosis.Referencing existed fault diagnosis theories and mechanical characteristics of verticalroller mill, the research on the algorithms of vertical roller mill fault diagnosis basedon expert system, decision tree, SVM, ANN and clustering analysis was carried out.And the simulations of algorithms have been implemented. (3) Based on the research of fault diagnosis algorithms of the vertical roller mill,a vertical roller mill condition monitoring and fault diagnosis system wasimplemented with the Java Web technology. The performance analysis and testing ofthe system was also carried out. The result of algorithms simulation and experimentsproved feasibility, accuracy and efficiency of these algorithms. Finally, the intelligentcondition monitoring and fault diagnosis was achieved through the vertical roller millfault diagnosis system.