Abstract: A simulation experiment based on EEG signal processing was designed for the automatic seizure detection application scenario, aiming to enhance students' understanding and application skills in basic machine learning principles. The experiment includes three stages: data cleaning, feature extraction, and seizure detection. In the data cleaning stage, a Butterworth filter is used for band-pass filtering, and the data is processed with overlapping slices. During the feature extraction stage, time-domain and frequency-domain features are calculated and combined into feature vectors. In the seizure detection stage, various classifiers, such as K-Nearest Neighbors, Support Vector Machines, and Random Forests, are employed for detection. Through this experiment, students can learn the processing methods of nonlinear signals, apply machine learning to solve practical problems, gain insight into real-world applications in biomedical engineering, and enhance their comprehensive practical skills, innovative thinking, and awareness of engineering ethics.