Large-scale equipment in different industries is becoming more and more electronic, complicated and integrated. In order to provide technical support for the normal operation of industrial equipment, fault diagnosis becomes extremely important. Intelligent fault diagnosis, as the name suggests, is to combine fault diagnosis and artificial intelligence theory to replace traditional fault diagnosis methods in a smarter way to obtain more accurate diagnosis results. Machine learning has become a current technical hotspot. In view of its advantages in efficiently extracting information and processing large amounts of data, more and more attention has been paid in the field of fault diagnosis. This paper systematically introduces the related concepts of machine learning and fault diagnosis, and summarizes fault diagnosis methods and applications based on support vector machines and neural networks. Finally, this paper makes a further outlook on the development of intelligent fault diagnosis technology based on machine learning.