Monitoring And Diagnosis System Of Centrifugal Compressor Based On Augmented Reality

Centrifugal compressor is a common equipment for compressing gas.Because of its compact structure,high speed and good reliability,it has been widely used in petrochemical industry.In order to know the state of centrifugal compressor in time and ensure its safe and stable operation,a monitoring and diagnosis system of centrifugal compressor based on augmented reality technology is developed.Through AR technology,the system presents the equipment information,monitoring parameters and diagnosis results of centrifugal compressor to the on-site inspection and maintenance personnel in a more real and vivid way,which improves the work efficiency of centrifugal compressor inspection.The basic structure and working principle of centrifugal compressor are introduced,and the common faults of centrifugal compressor are summarized and analyzed.The overall structure of the monitoring and diagnosis system of centrifugal compressor is designed,and the introduction and layout of the selected monitoring equipment are completed.The centrifugal compressor monitoring and diagnosis system based on augmented reality technology adopts unity 3D as the development engine,Vuforia as the main development kit,and completes the overall development combined with the monitoring and diagnosis system.The functions that the system can realize include the superposition of virtual model and real equipment,the scaling,rotation and de card of virtual model,viewing the monitoring data and diagnosis results of centrifugal compressor with SQLite database as the carrier and the explosion diagram and perspective diagram of centrifugal compressor.At present,the development of the system is basically completed,working normally and with high stability.The fault diagnosis algorithm of centrifugal compressor adopts the improved complete set empirical mode decomposition with adaptive noise and least squares support vector machine combined with particle swarm optimization.ICEEMDAN overcomes the problems of mode aliasing and false components of empirical mode decomposition and its improved algorithm.The signal is decomposed by ICEEMDAN to produce several IMF components.Select the larger components according to their correlation coefficients,calculate the sample entropy of these IMF components and input them into the classifier as characteristic parameters.PSO-LSSVM classification model is improved on the basis of least squares support vector machine algorithm.Two parameters are optimized by particle swarm optimization algorithm to improve the classification accuracy of the algorithm.Finally,the measured data of centrifugal compressor is brought into the fault diagnosis algorithm,and the diagnosis accuracy is 80.6%,which proves that the diagnosis process is suitable for centrifugal compressor and the effect is good.