With the ever-increasing population and the advancement of technology , high energy consumption is growing rapidly . As a result , use of alternative sources for fossil fuels is essential . One of the most common and abundant sources of energy is the wind . Wind turbines , as the most widely used system for generating electricity from the wind , are made up of many components and treated as a complex system . Similar to any other complex system , wind turbines are susceptible to different types of faults that if not diagnosed and treated properly could lead to safety hazards and substantial losses . Many methods have been proposed for fault diagnosis in the past decades . A category of methods that do not explicitly require a mathematical model of the system , are data based techniques . Support Vector Machine (SVM) and Neural Networks are two of the most popular of these methods . In this thesis , wind turbine fault diagnosis based on SVM is developed and compared to neural networks . The data required for this thesis is obtained from a benchmark wind turbine model which is widely used among researchers in this field . Five different faults , in two distinct operation regions of the benchmark wind turbine model , are considered . The output power of the wind turbine and the wind speed are collected and used for fault diagnosis . The collected data is broken into two parts which were used for training and validation of the proposed diagnosis framework . Using data from the training set an SVM model is trained to separate the faulty and normal states of the wind turbine operation . The trained model is then validated using the validation data . All these steps were carried out for a neural network based model as well and the results were compared with those obtained from the SVM model . Finally , a discussion on the results and some suggestions are given for future work. keywords: wind turbine, Fault detection, methods based on the data, SVM, Neural network