Determining the Working Behaviour of Hydraulic System Using Support Vector Machine

Abstract

Hydraulic systems acquire one of the most prominent places in modern industries. The use of hydraulics has increased with a very high rate, but at the same time, there is a lack of significant research in this particular field. As per the industry point of view, the working, as well as the proper monitoring of the hydraulic systems, is much desirable. In this manuscript, the primary focus is on the condition monitoring of the sub-system responsible for the proper functioning of the hydraulic system using artificial intelligence technique. Raw signals are captured from the pressure sensor followed by various statistical feature extractions. The correlation matrix and dependency plot have been used to visualize the inter-relationship between the extracted features. Extracted features are analysed and fed to the support vector machine (SVM) classifier. Tenfold cross-validation followed by grid search cross-validation has been employed to choose the superlative combination of hyperparameters for the classifier. The combination of chosen hyperparameters enhances the classification accuracy of the support vector machine. © 2021, Springer Nature Singapore Pte Ltd.