Abstract:Acoustic emission(AE)technique is an effective method for film seal face friction condition monitoring.However,it is difficult to separate the desired signal from AE original signal because of industrial background noise.To solve the problem,the selftuning Kalman filtering technique based on ARMA model was proposed to process the AE original signal.This filter converges to the steadystate optimal Kalman filter,even when model parameters and noise properties of system are unknown.Therefore,after filtering,the characteristics of AE signal are more prominent,and it is conducive to the detection of film seal face friction condition.RBF neural network was established,by using signal features in time domain,frequency domain and timefrequency domain as the input to training the network,the pattern identification of film seal face friction condition was realized.The experimental result shows that,this method can identify the friction condition effectively and in realtime,and the results of the pattern identification are consistent with the results of direct measurement by eddy current sensor.