Marine buoy data is an important foundation for marine scientific research, and researchers can use this data to analyze the balance of marine ecosystems, trends in climate change, and so on. However, outliers in ocean buoy data can interfere with the accuracy of research results. In view of this situation, this study proposes an outlier detection method for ocean buoy data based on depth support vector machine technology. First, the buoy data is collected through the sensor installed on the marine buoy and preprocessed; Secondly, data features are extracted to capture useful information related to outlier detection in the data; Finally, the depth support vector machine model is constructed to detect outliers in the buoy data. In testing, it was found that the method performed well in four aspects: accuracy, accuracy, recall and F1 score, which indicates that this method can more effectively and accurately identify the outliers in the ocean buoy data.