Research On Fault Diagnosis And Early Warning Of Power Plant Fans

The safe work of the three major fans of power plants(primary fan,blower,induced draft fan)are of great significance to the safety,economy of the unit.Due to the gradual shift of national policies to new energy generation,thermal power plants should participate in the deep peak shaving of the power grid,and the load of the unit will fluctuate frequently.;accordingly,various types of fans equipment of power plant are also often in a load operation changing state,which increases the possibility of equipment accidents and brings hidden dangers to the safety of the fan equipment.In view of the above problems,the fault diagnosis and early warning of fan is studied.Firstly,the technical route of multivariate state estimation technique(MSET)to model the state of power plant fans is studied.And the specific modeling variables are summarized.In the process of data preprocessing,the concept of the data band is proposed for the case where there are many abnormalities in the historical data,and the data band is constructed by using the historical data of the fan operation to achieve the purpose of cleaning the read data.Then the historical data is standardized,and then the memory matrix is constructed by the density peak clustering(DPC)to establish the MSET model of the normal state of the fan,and a fault warning process based on density peak clustering MSET algorithm is proposed.Then a primary fan is modeled and the model is verified by using the normal state history data of the fan.The results show that the established fan state model has high precision and can meet the requirements of power plant fan fault warning application.The threshold of fault warning is proposed based on historical normal operation data.Then,a fault data of the fan is used as an example to study the application.The results verify the effectiveness of the method,in this case,an alarm can be issued about 60 minutes before the power plant and the method can accurately realize the fault warning of the fan.Finally,combined with the data band and the variation characteristics of the fan parameters at the time of the fault,the judgment logic of the fan stall warning is proposed,and the fault instance is verified.The results show that the model can provide early warning of the fan stall,and in this case,an alarm can be issued about 215 seconds before the fault,which can make the operator have more time to adjust the fan and reduce the loss of stall.