The high power passing through transmission systems and the high costs due to the fault occurrence in these lines have encouraged researchers to pay special attention to protection issues in this area. The limitations and deficiencies of traditional protection methods and their strong dependencies on the system operating conditions doubles the importance of early fault detection and its prediction utilizing new techniques. Timely detection and warning issuance toward the possibility of fault occurrence can be accomplished by analyzing the data and information obtained from the system and examining the relationships between different parameters. In this paper, machine learning methods are used, which have the ability to predict the occurrence of faults with appropriate accuracy independent of the operating area of the system. To evaluate the performance of the models, a large amount of data has been generated in various operating conditions and applied as input to the algorithms under study. Also, the effects of different weather conditions as one of the important factors have been considered. For the sake of greater generality, accuracy check, and comparability of the results, three methods including KNN, SVM, and decision tree in two modes (unbalanced and balanced data in the existing classes) have been used, and the outcomes have been presented. The simulations and modeling presented in this paper have been implemented using Python and MATLAB.