Research on Efficient Verification and State Recognition Method for the Action Reliability of Manual Globe Valve

As a typical valve in primary system, manual globe valve is of great importance to maintain system operation and protect system safety. In order to verify the action reliability of the nuclear-grade manual globe valve and determine its operation state accurately and quantitatively, this paper studies and establishes an integrated intelligent operation device for manual globe valve action test, and proposes a method for identifying the state of the manual globe valve based on the combination of wavelet packet decomposition and support vector machine (SVM). Firstly, the torque signal is employed as the characteristic curve and the wavelet packet decomposition technique is utilized to extract the time-frequency domain features. The time domain and time-frequency domain features are integrated to construct the hybrid feature vector. Secondly, the Principal Component Analysis (PCA) is used to perform the dimensionality reduction analysis on the feature vectors to obtain fault feature vectors. Finally, the support vector machine (SVM) method is employed to identify the action state of valve. The results shows that the device constructed in this study solves the problems of long time-consuming and low efficiency in verifying the reliability of manual globe valve actions, as well as the difficulty in quantifying the evaluation of the action process. The proposed method can identify the three action states of the valve accurately and efficiently.