Machine learning-based tool health condition prediction of friction stir welding is major highlighting point in past few years in technology field. So, the continuous monitoring and predicting of tool life is most difficult task in industrial side, high rate of temperature generation and large amount of vibrations are willing to spoil the life of tool during welding process. In order to simplify or rectify these kinds of problems, vibration analysis-based machine learning approach was using. In this paper, AA5083 plates tend to weld through FSW by using tool steel. Among the all AA family, AA5083 has the weld ability and good hardness number. H13 tool steel is commonly using all friction stir welding operations. Tool steel possesses the high toughness, heat resistance, and wear resistance, and along with these three, tool steel possesses high hardness too. Data acquisition is major impact on condition monitoring technique. Piezoelectric transducer will help to extract the vibration signal from the rotating tool, and ADC in hardware will convert the signal to understanding form. Machine learning approach is one the recent technique in the world were continuous monitoring of a functioning object or machine component or rotating member in order to predict the life of component and need to make predictive maintenance of the machinery. There are different kind of approaches are available in machine learning. Support vector machine is the one, which will help in prediction of a performing object. Support vector machine is the powerful tool in both Digital signal processing is the major contributor in condition monitoring of machines in which signal was captured from the vibrating part through sensor called piezoelectric transducer theoretical analysis and practical prediction approach. Linear SVM, quadratic SVM, cubic SVM, fine Gaussian SVM, medium Gaussian SVM, and coarse Gaussian SVM are the sub-SVM algorithm that comes under support vector machine.