Statistic analysis and predication of crane condition parameters based on SVM

Abstract

Through statistic analysis of vibration and temperature signals of motor on the container crane hoisting mechanism in Waigaoqiao port, the feature vectors with vibration and temperature are obtained. Through data preprocessing and training data, Training models of condition parameters based on support vector machine (SVM) are established. The testing data of condition monitoring parameters can be predicted by these training models. During training the models, the penalty parameter and kernel function of model are optimized by cross validation. The research showed the predicted results of model using vibration and temperature is much better than the results only by vibration signal or temperature modeling.