Condition-based maintenance (CBM) involves continuously monitoring the status of an industrial device to determine whether it requires maintenance. It can particularly benefit highly complex devices like wind turbine generators (WTGs). Any critical conditions of such systems can be recognized via machine learning. This paper, therefore, aims to propose an approach to the CBM of WTGs based on the use of data from SCADA for training machine learning models. Such models were first tested to understand whether they could model the optimal behavior of a WTG. Next, using the models and a specific control chart, defined as sum-of-events, a visual analysis was carried out to determine any abnormal behavior. The results were decidedly interesting and are extensively described in the paper.