Research On Condition Monitoring And Fault Diagnosis Method Of Pulverizing System In Thermal Power Plant

The pulverizing system is an important auxiliary system in thermal power plants,and its safe and stable operation affects the reliable operation of the entire thermal power plant.The complex process flow of the equipment,the harsh working environment,and the long-term high-load operation can easily lead to various types of failures in the system.Therefore,it has important scientific research significance and engineering application value to study and establish an efficient pulverizing system condition monitoring and fault diagnosis system.This paper focuses on the medium-speed grinding positive pressure direct blowing pulverizing system,and the main contents include:(1)The structure,working principle and common fault types of the milling system are introduced.According to the actual parameter settings,the common faults of the milling system are divided into three categories,which are the faults that can be measured by a single measuring point,such as the oil pressure loaded by the mill.,the temperature of the mill outlet is abnormal,etc.;the failure of multi-point parameter coupling,such as coal blocking,coal breaking and spontaneous combustion of the coal mill,etc.;the failure of no measurement point parameters,such as the vibration failure of the coal mill.By analyzing the mechanism of various types of faults,the parameters for monitoring the operation status of the milling system are selected,which lays the foundation for the subsequent status monitoring and fault diagnosis research.(2)In view of the changeable operating conditions of the milling system,a model for monitoring the operation status of the milling system based on a deep bidirectional gated recurrent neural network was proposed.The bidirectional gated recurrent neural network was introduced into the stacked autoencoder to construct a deep bidirectional gated Recurrent neural network can fully extract the internal features of time series and improve the dynamic expression ability of the network.The sliding window method is used to construct the state monitoring model of the milling system,and the early warning threshold setting rules based on the Gaussian distribution theory are used for abnormal state warning.Through the verification of the actual data of the power plant,it is found that the established milling system state monitoring model can realize early fault warning,and Through the state monitoring of a single operating parameter,fault location can be achieved and the purpose of early fault diagnosis can be achieved.(3)Aiming at the problem that the fault features cannot be fully extracted and the cause of the vibration cannot be accurately identified when the vibration fault occurs in the medium-speed coal mill,a vibration fault diagnosis method of the medium-speed coal mill based on the combination of SSA-VMD-m RMR-LSSVM is proposed.Firstly,the decomposition level and penalty factor of VMD are determined by the Pearson correlation coefficient and envelope entropy combined with the gray wolf optimization algorithm.Secondly,the improved VMD algorithm is used to decompose the grinding current vibration signal to obtain several IMF components.Domain index and VMD energy entropy are used to construct multi-dimensional fault feature vector,and then the optimal feature subset of fault feature vector is determined by m RMR algorithm,and LSSVM is used for fault identification.The test results show that the fault features after feature selection can quickly and accurately identify the fault,which is convenient for on-site personnel to take corresponding measures.(4)Finally,based on the design tool of Py Qt5,the state monitoring and fault diagnosis platform of the milling system is developed.Through the design of the human-computer interaction interface,the entire process of condition monitoring and fault diagnosis is visualized.