LS-SVC based recognition method of the centrifugal pump cavitation intensity

For Least Squares Support Vector Classification (LS-SVC) has prominent advantages in selecting model, overcoming over-fitting and local minimum, and solving the problems of the nonlinear and high-dimensional pattern recognition and etc. by employing structural risk minimization criterion, the method which the centrifugal pump cavitations intensity are identified by using LS-SVC is proposed. It is found that waveform factor, peak factor, impulse factor, margin factor and kurtosis factor can be used as LS-SVC input which recognize five cavitations operating conditions and identify its intensity from weak to strong by simulating calculation successfully. The vibration of centrifugal pump and underwater acoustic signals was regarding as the cavitations feature in the experiment. Five working states, such as the normal condition, the pump lift declining 1%, 2%, and 3% respectively, and performance collapse, were distinguished through the method mentioned in the paper. Finally, compared with identify result of BP and RBF neural networks, the reorganization rate of the LS-SVC is the highest and the operation time is largely reduced.