In the past, the data intrusion detection methods of embedded components in power terminals only processed the data numerically, which led to low data quality and poor detection effect. Therefore, a data intrusion detection method for embedded components of power terminal based on machine learning is designed. The embedded components of power terminal are used to process and denoise the collected data, and their data characteristics are extracted and decomposed. By constructing an embedded component data intrusion detection system of power terminal and combining with neural network, intrusion detection is carried out. In the experimental test, the designed data intrusion detection method has better detection effect and can effectively improve the security protection ability of power system.